docs(superpowers): v0.20 sub-item 1 bugfix1/2/3 specs + plans + DOGFOOD.md

3-round dogfood-driven fix cycle 의 산출물: - bugfix1 (Bug #2/#3/#4): spec 964 line + plan 848 line - bugfix2 (Bug #6/#7, #8 falsified): spec 308 line + plan 388 line - bugfix3 (Bug #9/#10/#11/#13/#14, #12 falsified): spec 410 line + plan 1043 line - docs/DOGFOOD.md: 전방위 dogfood checklist 의 전체 (§0 environment ~ §13 reference corpus) 각 round 의 spec/plan 가 critic + verifier round 2 closure ACCEPT 후 frozen. dogfood-driven evidence 기반. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-28 01:21:34 +00:00
parent 9b44e27dfe
commit 46e99470eb
7 changed files with 4794 additions and 0 deletions
--- a/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix-plan.md
+++ b/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix-plan.md
@@ -0,0 +1,849 @@
+---
+title: v0.20.0 sub-item 1 bugfix — implementation plan
+created: 2026-05-27
+status: ACCEPT (round 2 closure — Phase B complete)
+target_version: 0.20.0 (PR #189 force-update)
+spec: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md
+contract_sections: ["§9 (chunker_version cascade)"]
+parent_plan: docs/superpowers/plans/2026-05-27-pdf-scanned-ocr-plan.md
+review_history:
+  - 2026-05-27 plan round 0 (opus, drafter) — 5 step group A-E, 18 sub-action
+  - 2026-05-27 plan round 1 critic (opus, thorough) — NEEDS_DISCUSSION, HIGH 1 + MEDIUM 2 + LOW 3 + NIT 1 (7 finding)
+  - 2026-05-27 plan round 1 verifier (opus, thorough) — NEEDS_DISCUSSION, HIGH 4 + MEDIUM 3 + LOW 4 + NIT 3 (14 finding)
+  - 2026-05-27 plan round 1c rewrite (opus, drafter) — 21 finding 모두 적용 (critic 7 + verifier 14). detail = §8 round 1c rewrite changelog
+  - 2026-05-27 plan round 2 closure critic (opus) — ACCEPT, 21/21 applied + 4 NIT cosmetic
+---
+
+# v0.20.0 sub-item 1 bugfix plan
+
+> ACCEPT 된 spec (`docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md`, 965 lines, round 2 closure) 의 step decomposition. 3 bug (#2 walker code limit / #3 chunk_id collision Critical / #4 F4 fixture Pages tree) 의 force-update path (PR #189 base branch `feat/pdf-scanned-ocr` 위에 fix commits stack). **5 step (Group A-E), 18 sub-action, 4 commit + 1 verify-only step (= 5 step total, 4 commit boundary).** spec §6 의 16-row acceptance 가 본 plan 의 §4 verifier checklist 로 1:1 mapping.
+
+## §0 Pre-flight + branch state
+
+- **Branch**: `feat/pdf-scanned-ocr` (PR #189 base, HEAD = `b4d9e60` "chore(release): bump version 0.19.0 → 0.20.0"). 사용자 메모리 `feedback_pr_workflow` 따라 force-update path — 같은 branch 위에 fix commits 추가 + PR #189 force-push.
+- **Working dir**: `/home/altair823/kebab`.
+- **Env 강제** (`~/.claude/CLAUDE.md` "Disk Layout — 루트 디스크 보호가 최우선"):
+  - `export CARGO_TARGET_DIR=/build/out/cargo-target/target` — XFS 4TB 전용 디스크 격리, repo root `target/` 생성 방지.
+  - `export RELEASE_BIN="${CARGO_TARGET_DIR:-target}/release/kebab"` — release binary alias (Step 5 dogfood 의 모든 acceptance command 에서 사용).
+  - `export TMPDIR=/build/cache/tmp` — 대용량 임시 파일 보호.
+- **Cargo build 직렬화** (memory `feedback_serial_build_only`):
+  - per-crate: `-j 4` default (예: `cargo test -p kebab-chunk -j 4`).
+  - workspace: `-j 1` 강제 (`cargo test --workspace -j 1`, `cargo clippy --workspace -j 1` — 18 integration-test binary 동시 link 시 OOM).
+- **`target/` clean policy** (memory `feedback_cargo_clean_policy`): `/build` XFS 4TB 분리라 routinely clean 금지. `df -h /build` 의 `Avail < 500G` OR `du -sh $CARGO_TARGET_DIR` > 200G 시에만 clean. Step 5 E1 의 first cargo invoke 직전 1 회 conditional check, 임계 미달 시 skip + commit body 안 "skipped cargo clean — /build avail X TB" 1줄 record.
+- **dogfood KB layout 가정** (Step 5 E3 prerequisite, critic round 1 H-1 closure): canonical config path = `/build/cache/tmp/v0.20-dogfood/config.toml` (in-place, KB dir 안). 외부 backup file `/build/cache/tmp/v0.20-dogfood-config.toml` 은 **존재 안 함** — 본 plan 의 모든 acceptance command 는 in-place config 기준. Step 5 E3 의 KB clean 은 **destructive `rm -rf` 금지**, config 보존 selective clean 사용 (E3 detail 참조). dogfood config canonical path 는 본 §0 의 한 곳에서만 정의 — Step 5 E3 의 command 가 이 path 참조.
+- **HOTFIXES.md / README / HANDOFF / ARCHITECTURE 영향**: Step 2 B4 가 HOTFIXES.md entry 추가 (Bug #3 second-iteration patch). 그 외 사용자 visible surface 변경 0 — README + HANDOFF + ARCHITECTURE 갱신 0 (CLI flag / wire schema / TUI key / config 추가 0; chunker_version bump 은 internal cascade 라 release notes 만).
+- **wire schema 변경 0** — `ingest_progress.v1` + `ingest_report.v1` 추가 field 0. V00X migration 0. `chunks` table DDL unchanged.
+- **frozen design contract 변경 0** — design §9 cascade rule 자체 변경 0 (rule 의 직접 적용으로 chunker_version 만 bump).
+- **workspace version bump 0** — v0.20.0 이 이미 cut (commit `b4d9e60`). 본 plan 은 같은 v0.20.0 안의 cumulative bugfix (PR #189 force-update). Step 5 E5 의 PR force-push 만, release tag 재컷 0.
+
+## §1 Plan overview + spec linkage
+
+Spec §3 (Bug #2) + §4 (Bug #3) + §5 (Bug #4) 의 fix design 을 atomic step 으로 decompose. 핵심 sequencing:
+
+1. **Bug #2 walker code limit fix** (Step 1) — `is_code_file` helper + walker conditional + unit test. spec §3.4 + §3.5 의 diff 그대로 적용. 1 commit.
+2. **Bug #3 chunk_id fix + chunker_version bump** (Step 2) — `chunk_page` return tuple 4-tuple 확장 + caller `per_chunk_hash` suffix 를 `segment_start` 로 변경 + `VERSION_LABEL` `"pdf-page-v1"` → `"pdf-page-v1.1"` bump + module doc 갱신 + HOTFIXES.md entry + unit regression test. spec §4.4 + §4.4.1 + §4.5 의 diff 그대로 적용. 1 commit.
+3. **Bug #3 integration test** (Step 3) — `crates/kebab-app/tests/` 안 multi-scanned PDF chunk_id collision-free integration test. spec §4.5.1 의 MockOcrEngine pre-condition 결정 (Option A share 또는 Option B inline) 이 executor 의 first sub-action. 1 commit.
+4. **Bug #4 F4 fixture re-generation** (Step 4) — `tests/fixtures/_synth/mojibake.py` 의 pikepdf-based rewrite + F4 fixture binary regenerate + parse-pdf 의 3 신규 invariant test. spec §5.4 + §5.5 + §5.6 의 diff 그대로 적용. 1 commit.
+5. **Workspace verify + commit + PR force-push** (Step 5) — cargo workspace test `-j 1` + clippy `-D warnings` + dogfood re-run (`/build/cache/tmp/v0.20-dogfood` isolated KB, qwen2.5vl:3b 의 Ollama endpoint `192.168.0.47:11434`) + PR #189 force-push. spec §6 16-row consolidated acceptance 가 본 step 의 verifier checklist.
+
+ordering invariant:
+
+- **Step 1 || Step 2 || Step 4 mutually independent**: 3 bug 의 fix 가 서로 다른 crate (`kebab-source-fs` / `kebab-chunk` / `tests/fixtures` + `kebab-parse-pdf`) 의 file path 에 한정 — 동시 진행 가능. 정합성 우선 → Step 1 → Step 2 → Step 4 sequential.
+- **Step 2 < Step 3**: integration test 가 `kebab-chunk` 의 fix 된 chunk_id 계산 path 위에 의존. Step 2 의 GREEN 이 prerequisite.
+- **Step 4 < Step 5 dogfood**: F4 fixture regeneration 의 결과 binary 가 dogfood 의 9 PDF 중 1 (mojibake) — Step 5 E3 dogfood 의 `block_count: 1` invariant 검증 prerequisite.
+- **Step 1-4 all < Step 5 workspace test**: workspace 전체 test 가 production code + test 의 final state 위에서만 의미.
+
+commit 단위는 logical group 1 commit (atomic) — §7 sequencing summary 의 5-commit table 따름. 사용자 memory `feedback_pr_workflow` (gitea-pr + 리뷰 루프) 따라 force-update 후 `gitea-pr-review` skill 의 review 루프 진입.
+
+---
+
+## §2 Step group structure (Group A-E)
+
+| Step | Group | 분류 | sub-action |
+|---:|---|---|---|
+| 1 | A | Bug #2 walker code limit fix | A1 `is_code_file` helper + A2 walker conditional + A3 unit test |
+| 2 | B | Bug #3 chunk_id collision fix + chunker_version bump | B1 `chunk_page` 4-tuple + B2 caller `per_chunk_hash` + B3 `VERSION_LABEL` bump + B4 module doc + HOTFIXES.md + B5 unit regression test |
+| 3 | C | Bug #3 multi-scanned PDF integration test | C1 MockOcrEngine share decision + C2 integration test (conditional) |
+| 4 | D | Bug #4 F4 fixture re-generation | D1 mojibake.py pikepdf rewrite + D2 fixture regenerate + commit + D3 parse-pdf 3 invariant test |
+| 5 | E | Workspace verify + commit + PR force-push | E1 cargo workspace test -j 1 + E2 clippy -D warnings + E3 dogfood re-run + E4 commit + E5 PR #189 force-push |
+
+---
+
+## §3 Per-step detail
+
+### Step 1 (Group A): Bug #2 walker code limit fix
+
+spec §3 의 Option A (code path only) — `is_oversized` 호출을 `is_code_file(path)` conditional 로 gate. PDF/image/markdown 의 size 는 parser 단계 자체 검증 (lopdf load_mem 256 KB+ 정상, image OCR 의 max_pixels self-cap).
+
+#### Sub-action A1 — `is_code_file` helper 추가
+
+- **Files affected**: `crates/kebab-source-fs/src/code_meta.rs` (line 129 `is_oversized` 함수 직후, 또는 `code_lang_for_path` 정의 직후).
+- **Action** (spec §3.4 diff 그대로):
+  ```rust
+  /// Returns true when `path`'s filename/extension is recognised as a code
+  /// file (per `code_lang_for_path`). Used by the walker to apply
+  /// `[ingest.code].max_file_bytes` / `max_file_lines` only to code files,
+  /// not to PDF/image/markdown (which have their own size controls in
+  /// their respective parsers).
+  pub(crate) fn is_code_file(path: &Path) -> bool {
+      code_lang_for_path(path).is_some()
+  }
+  ```
+- **Acceptance**:
+  - `grep -c "pub(crate) fn is_code_file" crates/kebab-source-fs/src/code_meta.rs` = **1**.
+  - `cargo build -p kebab-source-fs -j 4` green.
+
+#### Sub-action A2 — walker conditional size check
+
+- **Files affected**: `crates/kebab-source-fs/src/connector.rs:168-190` (현재 verified line range).
+- **Action** (spec §3.4 diff 그대로 — `is_oversized` 호출 앞에 `is_code_file` short-circuit):
+  ```diff
+  -            // Size-cap check (byte or line limit).
+  -            if crate::code_meta::is_oversized(
+  -                &abs_path,
+  -                self.max_file_bytes,
+  -                self.max_file_lines,
+  -            )
+  -            .unwrap_or(false)
+  +            // v0.20.0 sub-item 1 bugfix (#2): size-cap applies ONLY to
+  +            // code files. PDF/image/markdown bypass — their parsers
+  +            // have their own size controls. spec §3.3.
+  +            if crate::code_meta::is_code_file(&abs_path)
+  +                && crate::code_meta::is_oversized(
+  +                    &abs_path,
+  +                    self.max_file_bytes,
+  +                    self.max_file_lines,
+  +                )
+  +                .unwrap_or(false)
+              {
+                  fs_skips.skipped_size_exceeded =
+                      fs_skips.skipped_size_exceeded.saturating_add(1);
+                  ...
+                  tracing::debug!(
+                      path = %rel_path.display(),
+                      max_bytes = self.max_file_bytes,
+                      max_lines = self.max_file_lines,
+  -                    "skip: file exceeds size cap"
+  +                    "skip: code file exceeds size cap"
+                  );
+                  continue;
+              }
+  ```
+- **Acceptance**:
+  - `grep -nE "is_code_file\(&abs_path\)\s*$" crates/kebab-source-fs/src/connector.rs` ≥ **1**.
+  - `grep -c "skip: code file exceeds size cap" crates/kebab-source-fs/src/connector.rs` ≥ **1**.
+  - `cargo build -p kebab-source-fs -j 4` green.
+
+#### Sub-action A3 — Bug #2 unit test 추가
+
+- **Files affected**: `crates/kebab-source-fs/src/connector.rs` 의 기존 `#[cfg(test)] mod tests` (spec §3.5 "기존 test module 에 추가" 명시 — 새 file 아님).
+- **Action** (spec §3.5 의 `size_cap_skips_only_code_files` test body 그대로):
+  - 300 KB PDF / 300 KB markdown / 300 KB `big.rs` (3 file) tempdir 합성.
+  - `FsSourceConnector` (`max_file_bytes = 262_144`, `max_file_lines = 5_000`) 의 `scan_with_skips(&SourceScope::default())`.
+  - assertions:
+    - `paths.contains("paper.pdf")` (PDF walker pass).
+    - `paths.contains("notes.md")` (Markdown walker pass).
+    - `!paths.contains("big.rs")` (code file walker skip).
+    - `skips.skip_examples.size_exceeded` 안 `big.rs` 1 entry, `paper.pdf` 0 entry.
+  - cfg helper: 기존 test module 의 `cfg_with_size_cap(root, max_bytes, max_lines)` 패턴 재사용 (필요 시 helper 추가).
+- **Acceptance**:
+  - `cargo test -p kebab-source-fs size_cap_skips_only_code_files -j 4` green.
+  - 기존 `ingest_report_counts_oversized_files_by_bytes` (fixture `huge.rs`) + `ingest_report_size_cap_by_line_count` (fixture `longfile.rs`) regression 0 — fixture 명이 `.rs` 라 새 conditional 통과 (invariant preserved).
+  - `cargo test -p kebab-source-fs -j 4` 전체 green.
+
+#### Commit (Step 1 전체)
+
+```
+fix(source-fs): apply size limit only to code files; PDF/image/markdown bypass walker cap (Bug #2)
+
+- crates/kebab-source-fs/src/code_meta.rs: add pub(crate) fn is_code_file
+- crates/kebab-source-fs/src/connector.rs: walker conditional `is_code_file && is_oversized`
+- crates/kebab-source-fs/src/connector.rs mod tests: size_cap_skips_only_code_files unit test
+- spec: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md §3
+```
+
+---
+
+### Step 2 (Group B): Bug #3 chunk_id collision fix + chunker_version bump
+
+spec §4.3 의 Option A (segment boundary `start` 를 `per_chunk_hash` suffix 로). `chunk_page` return tuple 을 3-tuple `(actual_start, chunk_end, slice)` → 4-tuple `(segment_start, actual_start, chunk_end, slice)` 로 확장 + caller `per_chunk_hash` suffix 를 `segment_start` 로 변경. `VERSION_LABEL` `"pdf-page-v1"` → `"pdf-page-v1.1"` bump (spec §4.4.1 round 1c M-1 decision — explicit cascade audit trail).
+
+#### Sub-action B1 — `chunk_page` 4-tuple expansion
+
+- **Files affected**: `crates/kebab-chunk/src/pdf_page_v1.rs:200-204` (doc comment) + `:205-289` (signature line 205 → closing `}` line 289). 본 critic round 1 + verifier round 1 의 actual probe 결과 정정 (L-1).
+- **Action** (spec §4.4 diff 그대로):
+  - doc comment 갱신 — `(char_start, char_end, text_slice)` → `(segment_start, actual_start, chunk_end, text_slice)`:
+    ```rust
+    /// Split a single page's text into ordered chunks, each represented as
+    /// `(segment_start, actual_start, chunk_end, text_slice)`.
+    ///
+    /// - `segment_start` = pre-overlap segment boundary. Strictly increasing
+    ///   across the returned vec. Use this for chunk_id uniqueness suffixes.
+    /// - `actual_start` = post-overlap start char index. May collapse to a
+    ///   previous chunk's `actual_start` under aggressive overlap policy.
+    ///   Use this for `SourceSpan::Page::char_start`.
+    /// - `chunk_end` = chunk's end char index (exclusive).
+    fn chunk_page(text: &str, target_bytes: usize, overlap_bytes: usize)
+        -> Vec<(usize, usize, usize, String)>
+    ```
+  - early return: `vec![(0, n, text.to_string())]` → `vec![(0, 0, n, text.to_string())]`.
+  - loop body 의 push: `chunks.push((actual_start, chunk_end, slice))` → `chunks.push((start, actual_start, chunk_end, slice))`. (`start = bounds[seg_idx]` 는 이미 local var 로 존재 — line 245.)
+  - overlap walk 의 `let prev_min = prev.0` 가 기존 tuple 의 첫 field = post-fix tuple shape 에서는 `prev.1` (actual_start) — spec §4.4 의 invariant 보존 위해 변경:
+    ```diff
+    -        let actual_start = if let Some(prev) = chunks.last() {
+    -            let prev_min = prev.0;
+    +        let actual_start = if let Some(prev) = chunks.last() {
+    +            // prev tuple shape = (segment_start, actual_start, chunk_end, slice).
+    +            // overlap walk floor = previous chunk's actual_start (prev.1).
+    +            let prev_min = prev.1;
+                ...
+    ```
+- **Acceptance**:
+  - `grep -nE "fn chunk_page.*-> Vec<\(usize, usize, usize, String\)>" crates/kebab-chunk/src/pdf_page_v1.rs` = **1**.
+  - `grep -c "let prev_min = prev.1" crates/kebab-chunk/src/pdf_page_v1.rs` ≥ **1**.
+  - `cargo build -p kebab-chunk -j 4` green (caller B2 sub-action 동시 적용 후 red 해소).
+
+#### Sub-action B2 — caller `per_chunk_hash` suffix → `segment_start`
+
+- **Files affected**: `crates/kebab-chunk/src/pdf_page_v1.rs:149-186` (현재 verified — `chunk` method 의 `for (...) in chunk_page(...)` loop start line 149 → loop end line 186, verifier round 1 L-2 정정).
+- **Action** (spec §4.4 diff 그대로):
+  ```diff
+  -            for (char_start, char_end, slice) in
+  -                chunk_page(&p.text, target_bytes, overlap_bytes)
+  +            for (segment_start, char_start, char_end, slice) in
+  +                chunk_page(&p.text, target_bytes, overlap_bytes)
+               {
+                   ...
+                   let span = SourceSpan::Page {
+                       page: page_num,
+                       char_start: Some(char_start_u32),
+                       char_end: Some(char_end_u32),
+                   };
+                   let block_ids: Vec<BlockId> = vec![p.common.block_id.clone()];
+  -                // Per-chunk policy_hash variant prevents chunk_id
+  -                // collision when a page produces multiple chunks. See
+  -                // module docs for rationale.
+  -                let per_chunk_hash = format!("{base_policy_hash}#c{char_start}");
+  +                // v0.20.0 sub-item 1 bugfix (#3): per-chunk policy_hash
+  +                // variant uses `segment_start` (pre-overlap boundary,
+  +                // strictly increasing) instead of `char_start` (post-
+  +                // overlap, may collapse to prev_min). See module docs +
+  +                // spec §4.1 root cause + HOTFIXES.md 2026-05-27.
+  +                let per_chunk_hash = format!("{base_policy_hash}#c{segment_start}");
+                   let chunk_id =
+                       id_for_chunk(&doc.doc_id, &chunker_version, &block_ids, &per_chunk_hash);
+                   ...
+               }
+  ```
+  - `SourceSpan::Page.char_start` 는 여전히 post-overlap `char_start` (= `actual_start`) 보존 — citation locality semantic 유지.
+- **Acceptance** (verifier round 1 M-2: B2+B4 가 같은 logical commit 안 → grep 시점 = Step 2 commit time, 즉 post-B4):
+  - `grep -c "#c{segment_start}" crates/kebab-chunk/src/pdf_page_v1.rs` ≥ **1** (B2 단독 적용 시 = 1 call site; B4 module doc 적용 후 = 2 — B4 acceptance 가 ≥ 2 검증).
+  - `grep -c "#c{char_start}" crates/kebab-chunk/src/pdf_page_v1.rs` = **0** (call site + module doc 모두 segment_start 로 교체 — B2+B4 의 same-commit consolidated invariant).
+  - sub-action-by-sub-action 분리 검증 시 B2 단독 grep `#c{char_start}` 는 module doc line 56 의 literal 잔존으로 ≥ 1 — Step 2 commit boundary 도달 후 = 0 으로 확정.
+
+#### Sub-action B3 — `VERSION_LABEL` bump `"pdf-page-v1"` → `"pdf-page-v1.1"` + hardcoded literal 2 site 갱신
+
+- **Files affected** (verifier round 1 H-1 의 actual probe `grep -rn '"pdf-page-v1"' crates/ --include='*.rs'` 결과 2 site enumerate):
+  - `crates/kebab-chunk/src/pdf_page_v1.rs:67` (현재 verified — `const VERSION_LABEL: &str = "pdf-page-v1";`).
+  - `crates/kebab-app/tests/pdf_pipeline.rs:168` (현재 verified — `assert_eq!(pdf_item.chunker_version.as_ref().map(|c| c.0.as_str()), Some("pdf-page-v1"))` hard assertion, v1.1 bump 후 fail).
+  - `crates/kebab-app/tests/pdf_pipeline.rs:368` (현재 verified — error message string literal `"pdf-page-v1 emits 0 chunks for the empty page; total = 2"`, hard assertion 아니지만 stale 방지).
+- **Action** (spec §4.4.1 결정):
+  - **(a) primary const bump** (`crates/kebab-chunk/src/pdf_page_v1.rs:67`):
+    ```diff
+    -const VERSION_LABEL: &str = "pdf-page-v1";
+    +const VERSION_LABEL: &str = "pdf-page-v1.1";
+    ```
+    기존 test `chunker_version_is_pdf_page_v1` (pdf_page_v1.rs:374) 의 assertion 은 `VERSION_LABEL` const 인용 → 자동 갱신, test code 변경 불요.
+  - **(b) test assertion literal 갱신** (`crates/kebab-app/tests/pdf_pipeline.rs:168`, required):
+    ```diff
+    -        Some("pdf-page-v1")
+    +        Some("pdf-page-v1.1")
+    ```
+  - **(c) test error message literal 갱신** (`crates/kebab-app/tests/pdf_pipeline.rs:368`, recommended):
+    ```diff
+    -        "pdf-page-v1 emits 0 chunks for the empty page; total = 2"
+    +        "pdf-page-v1.1 emits 0 chunks for the empty page; total = 2"
+    ```
+- **Acceptance**:
+  - `grep -nE 'const VERSION_LABEL: &str = "pdf-page-v[0-9.]+";' crates/kebab-chunk/src/pdf_page_v1.rs` 결과 = `"pdf-page-v1.1"`.
+  - `cargo test -p kebab-chunk chunker_version_is_pdf_page_v1 -j 4` green (VERSION_LABEL 인용이라 자동 통과).
+  - `grep -rn '"pdf-page-v1"' crates/ --include='*.rs' | grep -v 'pdf-page-v1\.1'` = 결과 **0** (regex 의 false-positive 방지 — `pdf-page-v1.1` 의 substring `"pdf-page-v1"` 은 ".1" suffix 로 exclude). `grep -v` filter 후 line 0 이면 stale literal 잔존 0.
+  - `cargo test -p kebab-app pdf_pipeline -j 4` green (line 168 assertion 갱신 후).
+
+#### Sub-action B4 — module doc 갱신 + HOTFIXES.md entry
+
+- **Files affected**:
+  - `crates/kebab-chunk/src/pdf_page_v1.rs:47-60` (현재 verified — module doc `## chunk_id collision deviation` 단락).
+  - `tasks/HOTFIXES.md` (new dated entry append, 기존 entry 위치 — file 의 latest entry 가 `2026-05-26` 이므로 그 위에 `2026-05-27 — v0.20.0 sub-item 1` entry insert; 본 file 의 chronological pattern 따름).
+- **Action**:
+  - **(a) module doc** — spec §4.4 의 갱신본 그대로:
+    ```diff
+    -//! Workaround that doesn't change the §4.2 recipe: feed a per-chunk
+    -//! variant `format!("{base_policy_hash}#c{char_start}")` into the
+    -//! recipe's `policy_hash` slot (so distinct chunks distinguish via
+    -//! different policy_hash inputs), while storing the unmodified
+    -//! `base_policy_hash` in `Chunk.policy_hash` so the field still answers
+    -//! "what policy was active". Logged in `tasks/HOTFIXES.md`.
+    +//! Workaround that doesn't change the §4.2 recipe: feed a per-chunk
+    +//! variant `format!("{base_policy_hash}#c{segment_start}")` into the
+    +//! recipe's `policy_hash` slot. `segment_start` is the pre-overlap
+    +//! segment boundary, strictly increasing across the returned chunks
+    +//! even when the overlap walk collapses `actual_start` to a previous
+    +//! chunk's `prev_min`. Unmodified `base_policy_hash` is stored in
+    +//! `Chunk.policy_hash` so the field still answers "what policy was
+    +//! active". v1.1 second-iteration patch — logged in
+    +//! `tasks/HOTFIXES.md` (2026-05-27).
+    ```
+  - **(b) HOTFIXES.md entry** (spec §4.4 의 entry body 그대로):
+    ```markdown
+    ## 2026-05-27 — v0.20.0 sub-item 1: chunk_id `#c{char_start}` workaround collapses under aggressive overlap (Bug #3 second-iteration patch)
+
+    **Symptom**: F2 (1580 chars OCR, scanned_page2.pdf) ingest 시
+    `DocumentStore::put_chunks (pdf): sqlite error: UNIQUE constraint
+    failed: chunks.chunk_id: ... Error code 1555: A PRIMARY KEY constraint
+    failed`. `kebab v0.20.0` (commit `b4d9e60`) dogfood (qwen2.5vl:3b 의
+    `192.168.0.47:11434` Ollama endpoint, `/build/cache/tmp/v0.20-dogfood`
+    isolated KB) `--force-reingest` 마다 reproducible.
+
+    **Root cause**: `crates/kebab-chunk/src/pdf_page_v1.rs:170` 의
+    `per_chunk_hash = format!("{base_policy_hash}#c{char_start}")` 에서
+    `char_start` = post-overlap `actual_start`. line 266-281 의 overlap
+    walk 가 `prev_min` floor 까지만 back-walk 하므로 aggressive overlap
+    + 첫 segment 가 작은 page (F2 의 한국어 OCR text: 첫 ~10 char 안
+    sentence-end → segment_1 = [0, 30], segment_2 = [30, n], overlap_bytes
+    240 / chars=80 → segment_2 의 actual_start 가 prev_min=0 으로
+    collapse) → 두 chunk 의 `#c0` suffix identical → identical chunk_id →
+    `chunks` PRIMARY KEY violation.
+
+    **Fix** (spec §4.4): `chunk_page` return tuple 에 `segment_start`
+    추가 (3-tuple → 4-tuple `(segment_start, actual_start, chunk_end,
+    slice)`), caller `per_chunk_hash` 의 suffix 를 `segment_start` 로
+    변경. `segment_start` 는 `bounds[seg_idx]` (dedup 후 strictly
+    increasing) — overlap walk 와 무관하게 모든 chunk distinct. citation
+    locality 의 `SourceSpan::Page.char_start` 는 여전히 post-overlap
+    `actual_start` 유지.
+
+    **chunker_version cascade**: `pdf-page-v1` → `pdf-page-v1.1` bump
+    (spec §4.4.1 round 1c M-1 결정, design §9 cascade rule 의 직접 적용).
+    multi-chunk PDF page (pre-OCR 시점 `metro-korea.pdf` 의 21 block /
+    34 chunk 같은 정상 path) 의 chunk_id 가 변경 — explicit user-facing
+    audit trail 확보, store layer 의 자동 invalidation report. v0.20.0
+    force-update path 라 사용자 cost zero (어차피 fresh ingest).
+
+    **Amends**: spec `docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md`
+    §4.4. parent design §4.2 chunk_id recipe 자체 unchanged (workaround
+    layer 의 internal computation 만 변경). parent PR #189
+    (`feat/pdf-scanned-ocr`, force-update path).
+    ```
+- **Acceptance**:
+  - `grep -c "#c{segment_start}" crates/kebab-chunk/src/pdf_page_v1.rs` ≥ **2** (module doc + line 170 의 actual call).
+  - `grep -c "2026-05-27 — v0.20.0 sub-item 1: chunk_id" tasks/HOTFIXES.md` = **1**.
+
+#### Sub-action B5 — unit regression test `multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids`
+
+- **Files affected**: `crates/kebab-chunk/src/pdf_page_v1.rs` 의 `#[cfg(test)] mod tests` (현재 verified — `make_pdf_doc(&[&str])` + `default_policy(target, overlap)` helper 이미 존재, line 300-371).
+- **Action** (spec §4.5 의 test body 그대로):
+  ```rust
+  #[test]
+  fn multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids() {
+      // 한국어 OCR text 의 trigger shape: 10 char "가" + ". " + 500 char "나".
+      // → first segment [0, 12), second segment [12, n).
+      //   page_text byte_len = 10*3 + 2 + 500*3 = 1532 > target_bytes=1500
+      //   → multi-chunk. overlap_bytes = min(240, 750) = 240 chars=80
+      //   → second chunk 의 actual_start 가 prev_min=0 collapse → same `#c0`.
+      //
+      // default_policy(500, 80) — target_tokens=500 → target_bytes=500*3=1500
+      // (한국어 3byte/char 환산), overlap_tokens=80 → overlap_bytes=min(240, 750)=240.
+      // verifier round 1 L-3 보강.
+      let early_seg: String = std::iter::repeat('가').take(10).collect();
+      let tail: String = std::iter::repeat('나').take(500).collect();
+      let page_text = format!("{early_seg}. {tail}");
+
+      let doc = make_pdf_doc(&[&page_text]);
+      let policy = default_policy(500, 80);  // target=1500 byte, overlap=240 byte
+      let chunks = PdfPageV1Chunker.chunk(&doc, &policy).unwrap();
+
+      assert!(
+          chunks.len() >= 2,
+          "expected ≥2 chunks for {} byte page; got {}",
+          page_text.len(),
+          chunks.len()
+      );
+
+      let mut ids: Vec<&str> = chunks.iter().map(|c| c.chunk_id.0.as_str()).collect();
+      ids.sort_unstable();
+      let total = ids.len();
+      ids.dedup();
+      assert_eq!(
+          ids.len(),
+          total,
+          "all chunk_ids must be unique even when overlap walks actual_start back to prev_min"
+      );
+  }
+  ```
+- **Acceptance**:
+  - `cargo test -p kebab-chunk multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids -j 4` green.
+  - `cargo test -p kebab-chunk deterministic_chunk_ids_1000 -j 4` green (기존 determinism invariant 보존).
+  - `cargo test -p kebab-chunk overlap_clamped_when_overlap_exceeds_target -j 4` green (기존 overlap clamp invariant 보존).
+  - `cargo test -p kebab-chunk -j 4` 전체 green.
+
+#### Commit (Step 2 전체)
+
+```
+fix(chunk): chunk_id collision under aggressive overlap; bump pdf-page-v1 → pdf-page-v1.1 (Bug #3)
+
+- crates/kebab-chunk/src/pdf_page_v1.rs: chunk_page returns 4-tuple
+  (segment_start, actual_start, chunk_end, slice); caller per_chunk_hash
+  suffix uses segment_start (pre-overlap boundary, strictly increasing)
+  instead of char_start (post-overlap, may collapse to prev_min).
+- VERSION_LABEL "pdf-page-v1" → "pdf-page-v1.1" (design §9 cascade,
+  explicit user-facing audit trail).
+- module docs: workaround description updated to segment_start.
+- mod tests: multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids
+  regression pin.
+- tasks/HOTFIXES.md: 2026-05-27 entry (symptom F2 1580 char OCR,
+  intra-doc collision root cause, second-iteration patch rationale).
+- spec: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md §4
+```
+
+---
+
+### Step 3 (Group C): Bug #3 multi-scanned PDF integration test
+
+spec §4.5 + §4.5.1 — `kebab-app` integration 수준의 chunk_id collision-free regression. real Ollama 회피 위해 `OcrEngine` trait 의 MockOcrEngine. 기존 `crates/kebab-app/tests/pdf_ocr_apply.rs:20-45` 의 private MockOcrEngine 가 같은 crate 의 별 test binary 라 직접 import 불가 — executor 의 first sub-action 으로 share path 결정.
+
+#### Sub-action C1 — MockOcrEngine share decision (executor 의 dependency 확인 task)
+
+- **Files affected** (Option 별 분기):
+  - **Option A (share via `tests/common/`)** — verifier round 1 H-2 의 actual probe 결과 정정:
+    - `crates/kebab-app/tests/common/mod.rs` 는 **이미 존재** (172 line `TestEnv` infrastructure, `#![allow(dead_code)]` + `pub struct TestEnv` + `pub fn ingest_md` + `pub fn lexical_query` 등). action = **`pub mod mock_ocr;` 1줄 append** (mod.rs 신규 X).
+    - `crates/kebab-app/tests/common/mock_ocr.rs` (**신규** file, MockOcrEngine lift + per-page ctor).
+    - 기존 `pdf_ocr_apply.rs:20-45` 의 MockOcrEngine struct + impl 제거 + `mod common; use common::mock_ocr::MockOcrEngine;` import 추가 + ctor call site migration (M-3 참조).
+    - 신규 integration test 가 `mod common; use common::mock_ocr::MockOcrEngine;` 으로 share.
+  - **Option B (inline 중복)**: 신 test file `multi_scanned_pdf_ingest_no_chunk_id_collision.rs` 안에 inline `struct LocalMockOcr` + `impl OcrEngine for LocalMockOcr` (test isolation 우선, common/mod.rs touch X).
+- **Action**:
+  - **(a) dependency probe** — spec §4.5.1 의 결정 path 따름:
+    ```bash
+    grep -rn "impl OcrEngine" crates/kebab-parse-image/src/ crates/kebab-app/tests/
+    # 실 결과:
+    #   crates/kebab-parse-image/src/ocr.rs:235 — production OllamaVisionOcr.
+    #   crates/kebab-app/tests/pdf_ocr_apply.rs:25 — test-only MockOcrEngine.
+    ls crates/kebab-app/tests/common/mod.rs
+    # 실 결과: -rw-r--r-- ... 172 line (TestEnv infrastructure 이미 존재).
+    ```
+  - **(b) executor 결정**:
+    - 기존 MockOcrEngine 의 ctor 가 `MockOcrEngine { expected_text: String, fail: bool }` — per-page 다른 text 길이 지원 위해 ctor signature 확장 필요 (예: `expected_text: Vec<String>` + internal `Mutex<usize>` cursor). 확장이 trivial + 두 test 가 같은 crate → **Option A 권장**.
+    - Option A 시 `pdf_ocr_apply.rs` 의 MockOcrEngine ctor 호출 site (현재 실 verifier probe = **10 instantiation site** at lines 140, 170, 193, 210, 242, 284, 311, 334, 359, 399 — critic round 1 L-2 의 "9 → 10" off-by-1 정정. struct define line 21 제외) 가 새 ctor signature 로 migration — backward-compat 위해 두 ctor (`MockOcrEngine::single(text, fail)` + `MockOcrEngine::per_page(texts, fail)`) 제공. **mechanical migration**: 각 site 의 `MockOcrEngine { expected_text: <text>, fail: <bool> }` → `MockOcrEngine::single(<text>, <bool>)` (10 site × 1 line edit, verifier round 1 M-3 의 actual cost).
+    - Option B (inline) 는 sharing 비용 > test 격리 가치 시. 본 plan 의 first preference = Option A.
+  - **(c) 결정 결과 record**: result file (`.omc/reviews/2026-05-27-v0.20-bugfix-plan-drafter-r1c-result.md`) 의 closing summary 의 §6 open question 1 에 결정 path 기록 — Option A 시 sub-action C2 의 file edit = (existing) `common/mod.rs` append 1 line + (new) `common/mock_ocr.rs` + (modify) `pdf_ocr_apply.rs` + (new) `multi_scanned_pdf_ingest_no_chunk_id_collision.rs` = 4 file. Option B 시 1 new file 만.
+- **Acceptance**:
+  - probe grep 결과 ≥ 2 line (production + existing mock).
+  - probe ls 결과 — `common/mod.rs` existing 확인.
+  - executor 의 결정이 plan 의 §6 open question OQ-1 안에 명시.
+
+#### Sub-action C2 — integration test 작성 (conditional on C1 결정)
+
+- **Files affected** (Option A 채택 가정, verifier round 1 H-2 정정):
+  - `crates/kebab-app/tests/common/mod.rs` (**existing** 172 line — `pub mod mock_ocr;` 1줄 append 만).
+  - `crates/kebab-app/tests/common/mock_ocr.rs` (**신규** — MockOcrEngine lift + per-page ctor).
+  - `crates/kebab-app/tests/pdf_ocr_apply.rs:20-45` (기존 inline impl 제거 + `mod common; use common::mock_ocr::MockOcrEngine;` add — file head 의 mod declaration 1 줄 추가) + ctor call site 10 개 mechanical migration (M-3).
+  - `crates/kebab-app/tests/multi_scanned_pdf_ingest_no_chunk_id_collision.rs` (**신규**) — `mod common; use common::mock_ocr::MockOcrEngine;` import.
+- **Action** (spec §4.5 의 test body — 본 plan 의 sub-action 안 expanded):
+  - **fixture**: F1 (`scanned_page1.pdf`, 779 char OCR) + F2 (`scanned_page2.pdf`, 1580 char OCR) + 1 synthetic small-page PDF (300 char) — 3 scanned PDF.
+  - **MockOcrEngine ctor**: per-page text vec `["text for F1", "text for F2 의 1580 char string", "text for synthetic 300 char"]` + `fail: false`.
+  - **isolated KB**: `tempfile::tempdir()` + `Config::default()` 의 `data_dir` 만 override + workspace `[ingest.pdf].enabled = true`.
+  - **assertion path**:
+    1. `kebab_app::ingest_with_config_opts(&cfg, ...)` (facade) 호출.
+    2. `report.items.iter().filter(|i| i.kind == IngestItemKind::Error).count() == 0` — chunk_id collision 시 발생할 `ErrorKind::Storage` row 부재.
+    3. `store.get_chunks_count() == sum(per-PDF chunk_counts)` — DELETE+INSERT path 의 final row count.
+    4. `store.get_all_chunk_ids().iter().collect::<HashSet<_>>().len() == chunks_count` — chunk_id global uniqueness.
+  - **executor degradation path** (spec §4.5.1 conditional downgrade): 만약 Option A 의 share 가 비용/위험 크고 Option B 도 비현실적 (예: integration setup 의 ExtractContext / Facade wiring 가 본 sub-action scope 초과) → §6 row 7 의 acceptance 를 conditional downgrade — `kebab-chunk` 의 unit-level invariant (Step 2 B5) 만으로 Bug #3 의 core regression 핀 확보, integration 회피.
+- **Acceptance**:
+  - `cargo test -p kebab-app multi_scanned_pdf_ingest_no_chunk_id_collision -j 4` green.
+  - `cargo test -p kebab-app pdf_ocr_apply -j 4` green (existing test regression 0 — `MockOcrEngine { expected_text, fail }` literal struct construction 10 ctor site 가 `MockOcrEngine::single(text, fail)` 로 migration 후, critic round 1 L-2 actual count).
+  - downgrade path 시: result file + commit body 안 "§6 row 7 conditional skip — Bug #3 core regression = kebab-chunk unit B5" 1줄 record.
+
+#### Commit (Step 3 전체)
+
+```
+test(app): multi-scanned PDF chunk_id collision-free integration test (Bug #3 regression)
+
+- crates/kebab-app/tests/common/{mod,mock_ocr}.rs: MockOcrEngine lift
+  with per-page text ctor (shared by pdf_ocr_apply.rs + new test).
+- crates/kebab-app/tests/multi_scanned_pdf_ingest_no_chunk_id_collision.rs:
+  3 scanned PDF (F1 + F2 + synthetic 300char) ingest via mock OCR,
+  assert all chunk_ids globally unique + zero ErrorKind::Storage rows.
+- spec: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md §4.5
+```
+
+Option B (inline) 또는 conditional downgrade 채택 시 commit body 와 file list 그에 맞춰 조정.
+
+---
+
+### Step 4 (Group D): Bug #4 F4 fixture re-generation
+
+spec §5 — `tests/fixtures/_synth/mojibake.py` 의 byte-level `re.sub` + 수작업 startxref edit 를 pikepdf 의 proper PDF surgery (open + delete /ToUnicode + save 자동 xref regen) 로 교체. F4 fixture 자체 (`crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf`) regenerate + 3 신규 invariant test.
+
+#### Sub-action D1 — `tests/fixtures/_synth/mojibake.py` pikepdf rewrite
+
+- **Files affected**: `tests/fixtures/_synth/mojibake.py` (전체 rewrite — 기존 byte-edit 패턴 폐기).
+- **Action** (spec §5.4 의 body 그대로):
+  - Step 1: reportlab 으로 Type 0 (CID) font 사용 한국어 PDF 합성 (정상 ToUnicode CMap 포함).
+  - Step 2: pikepdf 로 open + 모든 dictionary 의 `/ToUnicode` entry 제거 + `pdf.save(allow_overwriting_input=True)` (xref 자동 regen).
+  - Step 3: invariant 검증 — `len(pdf.pages) == 1` + `b"/ToUnicode" not in dst.read_bytes()`.
+  - 실패 시 비-zero exit code + stderr message (Step 2 의 removed count = 0 → exit 2; Step 3 의 page count mismatch → exit 3; ToUnicode 잔존 → exit 4).
+- **Dep install** (executor 의 pre-action):
+  ```bash
+  pip install --cache-dir /build/cache/pip pikepdf reportlab
+  python -c "import pikepdf; import reportlab; print(pikepdf.__version__, reportlab.Version)"
+  # font availability probe (critic round 1 L-3) — mojibake.py 의 hardcode path.
+  test -f /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf \
+      || sudo apt-get install -y fonts-dejavu-core
+  ```
+  CI 환경 미반영 — fixture 자체를 commit 하므로 generation 은 1회성 (Step 4 D2 의 executor local). `tasks/HOTFIXES.md` 에 pikepdf install hint 만 1줄 추가 가능.
+- **Acceptance**:
+  - `grep -c "import pikepdf" tests/fixtures/_synth/mojibake.py` = **1**.
+  - `grep -c "re.sub" tests/fixtures/_synth/mojibake.py` = **0** (byte-edit 패턴 폐기 확인).
+  - `test -f /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf` exit 0 (font probe, critic round 1 L-3 fast failover signal).
+  - `python tests/fixtures/_synth/mojibake.py /tmp/mojibake_dryrun.pdf && echo OK` exit 0 + stderr 무.
+
+#### Sub-action D2 — F4 fixture binary regenerate + snapshot regen + commit
+
+- **Files affected** (verifier round 1 H-4 + critic round 1 M-2 의 actual probe `grep -rn 'fixtures/mojibake.pdf' crates/` 결과 2 consumer enumerate):
+  - `crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf` (regenerate).
+  - `crates/kebab-parse-pdf/tests/snapshots/vector_pdf_canonical.json` (**snapshot baseline file 자체** — delete + auto-regen). verifier round 1 H-4 의 actual probe `text_extractor_regression.rs:59-64` 의 hand-rolled `unwrap_or_else { write baseline }` 패턴.
+  - `crates/kebab-parse-pdf/tests/text_extractor_regression.rs` (existing test — 코드 자체 변경 0, snapshot regen path 만 trigger).
+  - `crates/kebab-parse-pdf/src/text_quality.rs:96` (verifier round 1 H-4 의 2번째 consumer — `let bytes = include_bytes!("../tests/fixtures/mojibake.pdf");` 의 unit test/doctest 가 fixture binary 변경 시 동시 verify, 코드 변경 0).
+- **Action**:
+  - **(a) regenerate command**:
+    ```bash
+    python tests/fixtures/_synth/mojibake.py \
+        crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf
+    ```
+  - **(b) regenerate 후 manual probe**:
+    ```bash
+    python -c "import pikepdf; pdf = pikepdf.open('crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf'); print(len(pdf.pages))"
+    # expected: 1
+    grep -c "/ToUnicode" crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf
+    # expected: 0 (binary grep — Pages dict 안 ToUnicode 부재)
+    ```
+  - **(c) snapshot baseline regen** (verifier round 1 H-4 + critic round 1 M-2 의 actual mechanic — OQ-2 closure):
+    - `text_extractor_regression.rs:59-64` 는 `let baseline = std::fs::read_to_string("tests/snapshots/vector_pdf_canonical.json").unwrap_or_else(|_| { std::fs::write(baseline_path, &actual).expect(...); actual.clone() })` 의 hand-rolled pattern (insta crate 사용 X).
+    - fixture binary 변경 → 다음 cargo test 시 `actual != baseline` → `assert_eq!` fail.
+    - executor 의 regen step:
+      ```bash
+      rm crates/kebab-parse-pdf/tests/snapshots/vector_pdf_canonical.json
+      cargo test -p kebab-parse-pdf vector_pdf_extract_byte_identical_to_baseline -j 4
+      # 1st run: snapshot file 부재 → unwrap_or_else write 패턴이 새 baseline 작성 → assert pass.
+      cargo test -p kebab-parse-pdf vector_pdf_extract_byte_identical_to_baseline -j 4
+      # 2nd run: 새 baseline 와 byte-identical → assert pass (regression invariant 확립).
+      ```
+    - OQ-2 closure — insta crate 미사용, cargo-insta CLI 불요. spec §5.6 의 "기존 `text_extractor_regression.rs` 의 F4 baseline 갱신" 의 actual mechanic 명문화.
+- **Acceptance**:
+  - `stat crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf` size > 0.
+  - `grep -c "/ToUnicode" crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf` = **0**.
+  - python 의 page count probe = `1`.
+  - `test -f crates/kebab-parse-pdf/tests/snapshots/vector_pdf_canonical.json` exit 0 (snapshot regen 후).
+  - `cargo test -p kebab-parse-pdf vector_pdf_extract_byte_identical_to_baseline -j 4` 2회 연속 green (regen + verify, H-4 mechanic).
+  - `cargo test -p kebab-parse-pdf -j 4` 전체 green — D3 의 신규 test + text_quality.rs:96 의 2번째 consumer 도 동시 verify.
+
+#### Sub-action D3 — parse-pdf 의 3 신규 invariant test
+
+- **Files affected** (verifier round 1 H-3 의 actual probe 결과 결정):
+  - actual `ls crates/kebab-parse-pdf/tests/` 결과: `common`, `extractor.rs`, `fixtures`, `ocr_e2e.rs`, `page_image.rs`, `snapshots`, `text_extractor_regression.rs` — plan round 0 의 primary candidate `text_extractor.rs` 는 **존재 안 함**.
+  - **결정**: `crates/kebab-parse-pdf/tests/text_extractor_regression.rs` append (F4 fixture consumer locality + D2 snapshot regen mechanic 와 same file). 3 신규 `#[test] fn` append.
+  - 대안 (executor 가 file size / cohesion 고려해 split 결정 시): 신규 `crates/kebab-parse-pdf/tests/mojibake_invariants.rs`. plan first preference = append to `text_extractor_regression.rs`.
+- **Action** (spec §5.5 의 3 test body 의 path 정정 — verifier round 1 H-3):
+  1. `mojibake_fixture_load_yields_one_page` — `let bytes = include_bytes!("fixtures/mojibake.pdf");` (integration test 는 이미 `crates/kebab-parse-pdf/tests/` root, `text_extractor_regression.rs:42` 의 canonical pattern 따름; spec §5.5 의 `"../tests/fixtures/mojibake.pdf"` 가 잘못 — `"fixtures/mojibake.pdf"` 직접). `lopdf::Document::load_mem(bytes).unwrap().get_pages().len() == 1`.
+  2. `mojibake_fixture_has_no_tounicode_cmap` — CWD-relative `std::fs::read("tests/fixtures/mojibake.pdf")` 위험 회피 (cargo test 의 CARGO_MANIFEST_DIR ≠ CWD 환경 가능): `let bytes = include_bytes!("fixtures/mojibake.pdf");` 사용. `bytes.windows(b"/ToUnicode".len()).filter(|w| *w == b"/ToUnicode").count() == 0`.
+  3. `pdf_text_extractor_on_mojibake_yields_one_block` — `let bytes = include_bytes!("fixtures/mojibake.pdf");` + PdfTextExtractor 의 `1 Block::Paragraph per page` invariant 검증, `canonical.blocks.len() == 1`, `scanned candidate` warning 또는 non-empty text. ExtractContext setup 의 actual body 는 executor 가 `text_extractor_regression.rs` 의 existing helper (있을 시) 또는 spec §5.5 의 placeholder 의 expansion.
+- **Acceptance**:
+  - `cargo test -p kebab-parse-pdf mojibake_fixture_load_yields_one_page -j 4` green.
+  - `cargo test -p kebab-parse-pdf mojibake_fixture_has_no_tounicode_cmap -j 4` green.
+  - `cargo test -p kebab-parse-pdf pdf_text_extractor_on_mojibake_yields_one_block -j 4` green.
+  - `cargo test -p kebab-parse-pdf -j 4` 전체 green.
+
+#### Commit (Step 4 전체)
+
+```
+fix(parse-pdf): F4 mojibake.pdf via pikepdf surgery; preserve 1-page invariant (Bug #4)
+
+- tests/fixtures/_synth/mojibake.py: full rewrite — replace byte-level
+  re.sub + manual startxref edit with pikepdf open+del+save (auto xref
+  regen). Type 0 font + ToUnicode strip via dictionary walk.
+- crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf: regenerate.
+- crates/kebab-parse-pdf/tests/text_extractor_regression.rs: append 3
+  invariant tests (lopdf 1-page / no ToUnicode marker / PdfTextExtractor
+  1-block) — verifier round 1 H-3 의 file path decision (same locality
+  with snapshot regen).
+- crates/kebab-parse-pdf/tests/snapshots/vector_pdf_canonical.json:
+  delete + auto-regen via 2-run cargo test (hand-rolled unwrap_or_else
+  pattern, verifier round 1 H-4).
+- spec: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md §5
+```
+
+verifier round 1 NIT-2 정정 — commit scope `test-fixture` → `parse-pdf` (crate name, conventional commit typical scope).
+
+---
+
+### Step 5 (Group E): Workspace verify + commit + PR #189 force-push
+
+spec §6 의 16-row consolidated acceptance 를 본 step 의 verifier checklist. 모든 acceptance command 가 scriptable.
+
+#### Sub-action E1 — cargo workspace test `-j 1`
+
+- **Files affected**: 변경 0 (verification only).
+- **Action**:
+  - **(a) conditional cargo clean** — memory `feedback_cargo_clean_policy` (verifier round 1 NIT-1 정정: TB vs GB unit 혼동 회피 위해 `-BG` 으로 GB unit 강제):
+    ```bash
+    # /build avail 을 GB 단위 정수로 직접 가져옴 (df -BG output 의 'G' suffix 만 strip).
+    AVAIL_GB=$(df -BG --output=avail /build | tail -1 | tr -d ' G')
+    # CARGO_TARGET_DIR 의 size 도 GB 단위 정수로 (du -BG output).
+    TARGET_GB=$(du -BG -s "${CARGO_TARGET_DIR:-target}" 2>/dev/null | awk '{print $1}' | tr -d 'G')
+    # /build avail < 500 GB OR target > 200 GB → clean
+    if [[ "${AVAIL_GB:-9999}" -lt 500 ]] || [[ "${TARGET_GB:-0}" -gt 200 ]]; then
+        cargo clean
+    fi
+    ```
+    임계 미달 시 skip + commit body / result file 안 1줄 record (예: "skipped cargo clean — /build avail ${AVAIL_GB}G, target ${TARGET_GB}G").
+  - **(b) workspace test**:
+    ```bash
+    cargo test --workspace --no-fail-fast -j 1 2>&1 | tail -100
+    ```
+  - tail 100 line + final summary "test result: ok. N passed; 0 failed" 확인.
+- **Acceptance**:
+  - exit code 0.
+  - stdout 의 "test result: ok" + "0 failed".
+  - spec §6 row 14 (workspace full test pass) 충족.
+
+#### Sub-action E2 — `cargo clippy --workspace -- -D warnings`
+
+- **Files affected**: 변경 0.
+- **Action**:
+  ```bash
+  cargo clippy --workspace --all-targets -j 1 -- -D warnings 2>&1 | tail -50
+  ```
+- **Acceptance**:
+  - exit code 0.
+  - "warning" 키워드 0 (or `-D warnings` 가 자동 error 화).
+  - spec §6 row 13 (workspace clippy clean) 충족.
+
+#### Sub-action E3 — dogfood re-run (Ollama qwen2.5vl:3b 환경)
+
+- **Files affected**: 변경 0. `/build/cache/tmp/v0.20-dogfood/` (isolated KB, 동일 dogfood 재사용).
+- **Action** (memory `feedback_pr_workflow` + `_external/` invariant 따름):
+  - **(a) release build**:
+    ```bash
+    cargo build --release -p kebab-cli -j 4 2>&1 | tail -10
+    "${CARGO_TARGET_DIR:-target}/release/kebab" --version
+    # expected: kebab 0.20.0
+    ```
+  - **(b) dogfood KB clean + re-ingest 9 PDF** (spec §1.1 의 dogfood 환경 동일).
+
+    canonical config path = `/build/cache/tmp/v0.20-dogfood/config.toml` (§0 가정). 외부 backup file `/build/cache/tmp/v0.20-dogfood-config.toml` 은 **존재 안 함** — critic round 1 H-1 의 actual probe 결과. 따라서 config 의 **자체 backup 후 clean + restore** path 사용 (destructive `rm -rf` 시 config 동시 삭제 방지):
+
+    ```bash
+    # Step A: config 의 임시 backup (KB clean 전 보존).
+    cp /build/cache/tmp/v0.20-dogfood/config.toml \
+        /build/cache/tmp/v0.20-dogfood-config.toml.bak
+
+    # Step B: KB 전체 clean (config 포함 destructive — backup 으로 보존됨).
+    rm -rf /build/cache/tmp/v0.20-dogfood/
+    mkdir -p /build/cache/tmp/v0.20-dogfood/
+
+    # Step C: backup 에서 config restore.
+    cp /build/cache/tmp/v0.20-dogfood-config.toml.bak \
+        /build/cache/tmp/v0.20-dogfood/config.toml
+
+    # config.toml 안 [ingest.pdf].enabled = true, ollama endpoint =
+    # http://192.168.0.47:11434, ocr_model = qwen2.5vl:3b
+
+    # Step D: ingest.
+    "$RELEASE_BIN" ingest --config /build/cache/tmp/v0.20-dogfood/config.toml \
+        --json --force-reingest 2>&1 | tee /build/cache/tmp/v0.20-dogfood-ingest.ndjson
+
+    # Step E (optional): backup file cleanup. 다음 dogfood iteration 의 redundant
+    # backup 누적 방지. config 자체는 v0.20-dogfood/config.toml 가 in-place
+    # canonical, .bak 은 transient.
+    rm /build/cache/tmp/v0.20-dogfood-config.toml.bak
+    ```
+
+    (대안 selective-delete 의 single-step path — config 보존 + 그 외 destructive):
+    ```bash
+    find /build/cache/tmp/v0.20-dogfood/ -mindepth 1 -not -name 'config.toml' \
+        -exec rm -rf {} +
+    ```
+    실 procedure 는 plain 5-step 의 명료성 우선 (executor 의 default).
+  - **(c) acceptance**:
+    - spec §6 row 3: 9 PDF 의 `skipped_size_exceeded == 0` for non-code (= 모두 0 — workspace 가 code 0).
+    - spec §6 row 8: F1 + F2 의 `kind != "Error"` (chunk_id collision 부재).
+    - spec §6 row 12: mojibake.pdf 의 ingest item `block_count: 1`.
+    - spec §6 row 15: 9 PDF 모두 ingest, `errors = 2` (encrypted only — pre-existing dogfood baseline 동일).
+  - **Ollama 미가용 시 fallback**: endpoint 가 unreachable 면 본 sub-action 의 partial skip 가능 — workspace test (E1) + clippy (E2) 의 unit/integration 수준 evidence 로 spec §6 row 1, 2, 4-7, 9-11, 13-14, 16 충족 + dogfood row 3, 8, 12, 15 skip 1줄 record (commit body + result file).
+- **Acceptance**:
+  - ingest report 의 ndjson 안 errors = 2 (encrypted only).
+  - F1/F2/mojibake 각각의 item line `kind` field 가 success path (= `"new"` 또는 `"unchanged"`, not `"Error"`).
+  - dogfood log path: `/build/cache/tmp/v0.20-dogfood-ingest.ndjson` (commit body 안 reference).
+
+#### Sub-action E4 — commit 점검 + 최종 organize
+
+- **Files affected**: 모든 step 의 누적 changes.
+- **Action**:
+  - `git status` + `git log --oneline b4d9e60..HEAD` — Step 1-4 의 4 commit + Step 5 의 verify-only commit 0 (verification 만, commit 없음).
+  - 만약 work-in-progress 잔존 file 있으면 reset.
+  - commit message 의 `Co-Authored-By:` line 점검 (CLAUDE.md gitea-pr workflow).
+- **Acceptance**:
+  - `git log --oneline b4d9e60..HEAD | wc -l` = **4** (Step 1-4 의 각 1 commit).
+  - `git status` 의 untracked + modified = 0.
+
+#### Sub-action E5 — PR #189 force-push
+
+- **Files affected**: remote ref `gitea/feat/pdf-scanned-ocr`.
+- **Action** (gitea-ops skill 의 직접 호출 가능):
+  ```bash
+  git push gitea feat/pdf-scanned-ocr --force-with-lease
+  ```
+  - `--force-with-lease` — local 의 fetch state 와 remote HEAD 가 match 시에만 force-push (다른 collaborator 의 push 보호; 본 single-user 환경 cheap safety).
+  - PR #189 의 body 갱신 — Bug #2/#3/#4 fix summary + dogfood evidence 추가 (gitea API `PATCH /repos/altair823-org/kebab/pulls/189`).
+  - 사용자 memory `feedback_pr_workflow` 따라 `gitea-pr-review` skill 의 review 루프 진입 (multi-round critic + verifier).
+- **Acceptance**:
+  - `gh-equivalent` (gitea-ops `gitea-pr-status 189`) 의 head SHA = local `git rev-parse HEAD`.
+  - PR #189 의 commit count = 이전 force-push 시점 의 commit count + 4.
+  - sequencing summary 의 5-commit table (§7) 와 final state 일치.
+
+#### Commit (Step 5)
+
+verification only — git commit 0. Step 1-4 의 4 commit 가 final tree.
+
+---
+
+## §4 Verifier checklist (spec §6 16-row 1:1 mapping)
+
+각 row 가 scriptable command. step 5 E1-E3 의 누적 실행으로 모두 가능.
+
+| # | Verifier | Bug | step | 명령 |
+|---|---------|-----|------|------|
+| 1 | walker bypasses size cap for PDF | #2 | A3 | `cargo test -p kebab-source-fs size_cap_skips_only_code_files -j 4` |
+| 2 | walker still skips oversized code files | #2 | A3 | `cargo test -p kebab-source-fs ingest_report_counts_oversized_files_by_bytes -j 4` |
+| 3 | 256KB+ PDF/markdown ingest default config | #2 | E3 | dogfood: `$RELEASE_BIN ingest ...` 의 ingest report 의 `skipped_size_exceeded = 0` for non-code |
+| 4 | chunker collision regression test | #3 | B5 | `cargo test -p kebab-chunk multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids -j 4` |
+| 5 | chunker determinism preserved | #3 | B5 | `cargo test -p kebab-chunk deterministic_chunk_ids_1000 -j 4` |
+| 6 | chunker overlap clamp preserved | #3 | B5 | `cargo test -p kebab-chunk overlap_clamped_when_overlap_exceeds_target -j 4` |
+| 7 | integration: multi-scanned PDF ingest (conditional, §4.5.1) | #3 | C2 | `cargo test -p kebab-app multi_scanned_pdf_ingest_no_chunk_id_collision -j 4` (Option A/B downgrade path 시 skip + record) |
+| 8 | dogfood: F1 + F2 force-reingest errors=0 | #3 | E3 | dogfood: `$RELEASE_BIN ingest --force-reingest ...` 의 errors = 0 (encrypted 제외) |
+| 9 | F4 fixture lopdf 1-page invariant | #4 | D3 | `cargo test -p kebab-parse-pdf mojibake_fixture_load_yields_one_page -j 4` |
+| 10 | F4 fixture ToUnicode 부재 invariant | #4 | D3 | `cargo test -p kebab-parse-pdf mojibake_fixture_has_no_tounicode_cmap -j 4` |
+| 11 | F4 PdfTextExtractor 1-block invariant | #4 | D3 | `cargo test -p kebab-parse-pdf pdf_text_extractor_on_mojibake_yields_one_block -j 4` |
+| 12 | dogfood: F4 ingest block_count=1 | #4 | E3 | dogfood: mojibake.pdf 의 ingest item `block_count: 1` |
+| 13 | workspace clippy clean | all | E2 | `cargo clippy --workspace --all-targets -j 1 -- -D warnings` |
+| 14 | workspace full test pass | all | E1 | `cargo test --workspace --no-fail-fast -j 1` |
+| 15 | dogfood end-to-end 9 PDF | all | E3 | dogfood: 9 PDF 모두 ingest, errors = 2 (encrypted only) |
+| 16 | chunker_version cascade final value | #3 | B3 | `grep -nE 'pdf-page-v[0-9.]+' crates/kebab-chunk/src/pdf_page_v1.rs` 결과가 `"pdf-page-v1.1"` |
+
+executor 의 final step (E1-E3) 에서 16 row 모두 scriptable 실행 + result file 안 row-by-row pass/fail/skip 기록.
+
+#### Workspace baseline expected test count delta (verifier round 1 M-1 closure)
+
+`cargo test --workspace -j 1` (Step 5 E1) 의 expected `test result: ok. N passed` 의 delta 산수 — pre-fix baseline 대비:
+
+| Step | Sub-action | new test name | crate | type |
+|---|---|---|---|---|
+| 1 | A3 | `size_cap_skips_only_code_files` | kebab-source-fs | unit (in `mod tests`) |
+| 2 | B5 | `multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids` | kebab-chunk | unit (in `mod tests`) |
+| 3 | C2 | `multi_scanned_pdf_ingest_no_chunk_id_collision` | kebab-app | integration (new test binary) |
+| 4 | D3 | `mojibake_fixture_load_yields_one_page` | kebab-parse-pdf | unit-style integration |
+| 4 | D3 | `mojibake_fixture_has_no_tounicode_cmap` | kebab-parse-pdf | unit-style integration |
+| 4 | D3 | `pdf_text_extractor_on_mojibake_yields_one_block` | kebab-parse-pdf | unit-style integration |
+
+- **Option A (full path, C2 active)**: total = **+6 unit/integration test cases + 1 new integration test binary**.
+- **Option B (C2 conditional downgrade per §4.5.1)**: total = **+5 test cases + 0 new binary**.
+
+기타: B3 의 `chunker_version_is_pdf_page_v1` 기존 test 의 assertion content 변경 없음 (VERSION_LABEL const 인용) → test count delta 0. D2 의 `vector_pdf_extract_byte_identical_to_baseline` 기존 test 의 assertion 결과만 변경 (fixture 변경 → baseline 변경) → test count delta 0, snapshot regen action 만 추가.
+
+executor 가 E1 acceptance 의 N 비교 시 본 delta 산수 와 일치 확인 (regression 시 detection).
+
+---
+
+## §5 Risks (plan 단계)
+
+- **R-1 (MockOcrEngine sharing complexity)**: spec §4.5.1 의 Option A (`tests/common/mock_ocr.rs` lift) 가 기존 `pdf_ocr_apply.rs:20-45` 의 9 test 의 ctor migration 필요 — backward-compat ctor 2 개 (single + per_page) 도입 시 trivial, 실패 시 Option B (inline) downgrade. spec §6 row 7 conditional skip 가능.
+- **R-2 (chunker_version bump cascade scope)**: `pdf-page-v1.1` 의 영향 = multi-chunk PDF page 의 chunk_id 변경. parser_version / embedding_version / prompt_template_version / index_version unchanged — `kebab-eval::eval_runs.config_snapshot_json` 의 5-version snapshot 의 chunker_version field 만 새 값. parent design §9 의 cascade rule invariant 보존, eval baseline 의 re-run 권장 (spec §7.1 Risk 1 의 user-facing note).
+- **R-3 (F4 fixture binary churn)**: pikepdf 의 save output 가 reportlab+byte-edit 와 다른 PDF object ordering → SHA256 변경 + git binary diff noise. `text_extractor_regression.rs` baseline 도 새 fixture 의 actual output 으로 same-commit update — Step 4 D2 안 동시 처리.
+- **R-4 (dogfood Ollama 의존)**: spec §6 row 3 + 8 + 12 + 15 dogfood acceptance 가 real `192.168.0.47:11434` qwen2.5vl:3b 호출. endpoint 미가용 시 unit/integration evidence (row 1-2, 4-7, 9-11, 13-14, 16) 로 partial closure + commit body / result file 안 skip record.
+- **R-5 (pikepdf dependency install)**: Step 4 D1 의 mojibake.py 의 `import pikepdf` — 본 머신의 Python venv 에 pip install 필요. CI 의존성 미발생 (fixture commit 후 1회성 generation).
+- **R-6 (parent plan 와의 동시 진행 충돌 0 확인)**: parent plan (`2026-05-27-pdf-scanned-ocr-plan.md` round 1c ACCEPT) 의 Step 11 (final verify + PR open) 가 이미 commit `b4d9e60` 으로 closed. 본 plan 의 fix commits 가 그 commit 위에 stack — branch ordering 충돌 0.
+
+---
+
+## §6 Open questions deferred to executor
+
+- **OQ-1 (MockOcrEngine sharing path)**: spec §4.5.1 의 Option A (`tests/common/mock_ocr.rs` lift) vs Option B (inline) 결정. executor 의 Step 3 C1 안 first action — probe `grep -rn "impl OcrEngine"` 후 결정 + result file 안 record. plan first preference = Option A.
+- **OQ-2 (F4 baseline snapshot update tool)**: ✅ **CLOSED (round 1c, critic M-2 + verifier H-4)** — `text_extractor_regression.rs:59-64` 의 actual pattern = hand-rolled `unwrap_or_else { write baseline }` (insta crate 사용 X). regen procedure = snapshot file `tests/snapshots/vector_pdf_canonical.json` 삭제 + cargo test 2회 (1st auto-regen, 2nd verify). cargo-insta CLI 불요. detail = §3 Step 4 D2 의 Action (c).
+- **OQ-3 (pikepdf install command)**: `pip install` 의 cache-dir + venv 결정 — global `--user` pip 또는 fixture generation 전용 venv 또는 conda environment. plan 의 default = `pip install --cache-dir /build/cache/pip pikepdf reportlab` (memory `feedback_disk_layout`).
+- **OQ-4 (dogfood config.toml 의 endpoint 변경 시점)**: 본 dogfood 환경의 `192.168.0.47:11434` Ollama endpoint 가 변경되면 executor 가 alternative endpoint (`localhost:11434` 등) 로 override + result file 안 record.
+- **OQ-5 (PR #189 review 루프의 round 수)**: memory `feedback_pr_workflow` 의 gitea-pr + 리뷰 루프 — round 1 critic + verifier 의 결과에 따라 round 2/2c 진입 가능. 본 plan 은 round 0 (drafter) — review round 의 outcome 은 plan 외 scope.
+
+---
+
+## §7 Sequencing summary (logical commit boundaries)
+
+| commit # | step range | logical scope | file count |
+|---:|---|---|---:|
+| 1 | Step 1 (A1+A2+A3) | `fix(source-fs): apply size limit only to code files; PDF/image/markdown bypass walker cap (Bug #2)` | 2 |
+| 2 | Step 2 (B1+B2+B3+B4+B5) | `fix(chunk): chunk_id collision under aggressive overlap; bump pdf-page-v1 → pdf-page-v1.1 (Bug #3)` | 4 (pdf_page_v1.rs + HOTFIXES.md + pdf_pipeline.rs:168 + :368, verifier H-1) |
+| 3 | Step 3 (C1+C2) | `test(app): multi-scanned PDF chunk_id collision-free integration test (Bug #3 regression)` | **4 (Option A: existing common/mod.rs append + new common/mock_ocr.rs + modify pdf_ocr_apply.rs + new multi_scanned_pdf_ingest_no_chunk_id_collision.rs, verifier H-2)** / 1 (Option B) |
+| 4 | Step 4 (D1+D2+D3) | `fix(parse-pdf): F4 mojibake.pdf via pikepdf surgery; preserve 1-page invariant (Bug #4)` | **5 (mojibake.py + fixtures/mojibake.pdf + snapshots/vector_pdf_canonical.json + text_extractor_regression.rs (D3 append) + src/text_quality.rs:96 consumer verify, verifier H-4 + H-3 + NIT-2)** |
+| 5 | Step 5 (E1-E5) | verification only — git commit 0; final state = commits 1-4 위 PR #189 force-push | 0 |
+
+총 4 commit + 1 verify-only step. force-push 후 PR #189 의 head = local HEAD.
+
+---
+
+## §8 Round 1c rewrite changelog (drafter trace)
+
+round 1 critic + verifier 의 합산 21 finding (critic 7 + verifier 14) 적용. detail 은 result file (`.omc/reviews/2026-05-27-v0.20-bugfix-plan-drafter-r1c-result.md`) 의 §1 traceability matrix 참조. 본 §8 은 plan body 의 substantive change summary.
+
+### Critic r1 (7 finding)
+
+| ID | Severity | Action | Plan section |
+|---|---|---|---|
+| critic H-1 | HIGH | E3 dogfood config 의 backup 후 clean + restore 5-step procedure (외부 backup file 부재 reality 반영) | §3 Step 5 E3 (b) |
+| critic M-1 | MEDIUM | line 15 "17 sub-action" → "18 sub-action" | §0 prelude line |
+| critic M-2 | MEDIUM | D2 snapshot baseline 갱신 mechanic 명문 (hand-rolled `unwrap_or_else` pattern, OQ-2 closure) | §3 Step 4 D2 + §6 OQ-2 |
+| critic L-1 | LOW | B1 line range "200-289" → "200-204 (doc) + 205-289 (body)" 명시 | §3 Step 2 B1 |
+| critic L-2 | LOW | MockOcrEngine ctor count "9 test (existing)" → "10 instantiation site" (actual probe) | §3 Step 3 C1 + C2 |
+| critic L-3 | LOW | D1 pre-action 에 DejaVuSans.ttf existence probe 1줄 추가 | §3 Step 4 D1 |
+| critic NIT-1 | NIT | "5 logical commit" → "4 commit + 1 verify-only step (= 5 step total, 4 commit boundary)" | §0 prelude line |
+
+### Verifier r1 (14 finding)
+
+| ID | Severity | Action | Plan section |
+|---|---|---|---|
+| verifier H-1 | HIGH | B3 sub-action 에 `pdf_pipeline.rs:168` (hard assertion) + `:368` (error message) literal 갱신 명시 + acceptance grep regex 정밀화 (`grep -v 'pdf-page-v1\.1'`) | §3 Step 2 B3 |
+| verifier H-2 | HIGH | Step 3 Option A 의 `common/mod.rs` 가 existing infrastructure 반영 — `pub mod mock_ocr;` 1줄 append + 신규 `common/mock_ocr.rs` + `pdf_ocr_apply.rs` lift + 신규 integration test = 4 file edit | §3 Step 3 C1 + C2 + §7 commit 3 file count |
+| verifier H-3 | HIGH | D3 file path `text_extractor.rs` 부재 정정 → `text_extractor_regression.rs` append (locality with D2 snapshot regen). `include_bytes!` path 도 `../tests/fixtures/...` → `fixtures/...` 직접 + CWD-relative `std::fs::read` 회피 | §3 Step 4 D3 |
+| verifier H-4 | HIGH | D2 snapshot regen mechanic — snapshot file `tests/snapshots/vector_pdf_canonical.json` 삭제 + cargo test 2회 (1st auto-regen, 2nd verify) + `src/text_quality.rs:96` 2번째 consumer enumerate | §3 Step 4 D2 + §7 commit 4 file count |
+| verifier M-1 | MEDIUM | §4 verifier checklist 뒤에 expected workspace test count delta 산수 표 추가 (+6 unit + 1 integration, Option A / +5 + 0, Option B) | §4 (sub-section) |
+| verifier M-2 | MEDIUM | B2 acceptance phrasing 갱신 — "Step 2 commit time" 명시 + sub-action 별 grep 시점 명문 | §3 Step 2 B2 acceptance |
+| verifier M-3 | MEDIUM | C2 Option A 의 "기존 10 ctor site mechanical migration" 명령 명시 | §3 Step 3 C1 (b) |
+| verifier L-1 | LOW | pdf_page_v1.rs line range 200-289 → 205-289 (critic L-1 와 same edit pass) | §3 Step 2 B1 |
+| verifier L-2 | LOW | caller line range 155-185 → 149-186 | §3 Step 2 B2 |
+| verifier L-3 | LOW | B5 test scenario comment 의 target=1500 byte + overlap=240 byte 산수 보강 | §3 Step 2 B5 |
+| verifier L-4 | LOW | `ingest_pdf_ocr_smoke.rs` 의 grep B3 scope safety 확인 (별도 action 0, finding 자체 = no action) | (verified safe) |
+| verifier NIT-1 | NIT | E1 의 `df -h` unit 처리 산수 정밀화 → `df -BG --output=avail` 으로 GB unit 강제 | §3 Step 5 E1 (a) |
+| verifier NIT-2 | NIT | Step 4 commit scope `test-fixture` → `parse-pdf` (crate name) | §3 Step 4 commit |
+| verifier NIT-3 | NIT | dogfood config canonical path 의 single-definition (in §0) + 모든 acceptance command 참조 | §0 pre-flight + §3 Step 5 E3 |
+
+### Summary
+
+- frontmatter `status` `draft (round 0)` → `draft (round 1c)`.
+- frontmatter `review_history` 에 round 1 critic + verifier + round 1c rewrite 항목 3 줄 add.
+- plan body line 15 의 prelude statement 2 token 정정 (sub-action count + commit boundary 표현).
+- §0 pre-flight 에 dogfood KB layout 가정 1 bullet add.
+- §3 5 step 의 sub-action body 의 detail 보강 (file path / acceptance grep / mechanic / migration cost).
+- §4 verifier checklist 의 expected test count delta sub-section add.
+- §6 OQ-2 closure 표시 (✅ CLOSED).
+- §7 sequencing summary 의 file count 갱신 (commit 2: 2→4, commit 3: 3-4→4, commit 4: 3-4→5).
+- §8 round 1c rewrite changelog (본 단락) populate.
+
+총 plan body line 변경 = ~+250 net add (round 0 698 line → round 1c ~950 line).
--- a/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix2-plan.md
+++ b/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix2-plan.md
@@ -0,0 +1,388 @@
+---
+title: "v0.20.0 sub-item 1 bugfix round 2 — plan"
+created: 2026-05-27
+status: "DRAFT round 0"
+spec_path: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md
+spec_status: ACCEPT (round 1c, 308 line)
+critic_round_1: .omc/reviews/2026-05-27-v0.20-bugfix2-spec-critic-r1-result.md
+critic_round_2: .omc/reviews/2026-05-27-v0.20-bugfix2-spec-critic-r2-result.md
+step_count: 4 (3 commit + 1 sanity-only)
+commit_count: 3
+branch: feat/pdf-scanned-ocr
+head_at_draft: e674ff4
+---
+
+# v0.20.0 sub-item 1 bugfix round 2 — plan
+
+## §0 Overview
+
+Spec ACCEPT (round 1c, 9/9 critic finding 반영) 의 implementation map. fix scope = 2 bug:
+
+- **Bug #6** (Critical): `?Identity-H Unimplemented?` mojibake marker bypass — `crates/kebab-parse-pdf/src/text_quality.rs::compute_valid_char_ratio()` 의 marker strip + dominance heuristic.
+- **Bug #7** (Minor doc): `kebab search --help` 의 `--media` value list 에서 `code` 누락 — clap doc-comment + SKILL.md 동기 갱신 + CLI help regression test.
+
+총 변경 file 5 + 신규 test file 1 + HOTFIXES entry. branch = `feat/pdf-scanned-ocr` (HEAD = `e674ff4`, round 1 의 4 commit 적층 위에 round 2 의 3 commit append). env = `CARGO_TARGET_DIR=/build/out/cargo-target/target`. fresh release binary = `/build/out/cargo-target/target/release/kebab`.
+
+**non-scope (spec §2.2 + 본 plan §5 OQ-4)**: spec ACCEPT 의 surface 외에 `crates/kebab-mcp/src/tools/search.rs:44`, `crates/kebab-core/src/search.rs:32+52`, `crates/kebab-app/src/ingest_progress.rs:69`, `crates/kebab-cli/tests/wire_schema_breakdowns.rs:35` 가 같은 stale value list (`markdown, pdf, image, audio, other` — `code` 누락) 을 보유. spec 의 frozen grep boundary (`integrations/` + `crates/kebab-cli/src` + `docs/wire-schema/v1`) 외이므로 본 round 의 commit 대상 X — follow-up issue 권장.
+
+---
+
+## §1 Step table
+
+| Step | Title | Scope summary | Commit subject | Files touched |
+|------|-------|----------------|----------------|----------------|
+| 1 | Bug #6 implementation | `MOJIBAKE_MARKERS` const + `compute_valid_char_ratio()` rewrite + 2 new unit test + HOTFIXES entry | `fix(parse-pdf): strip Identity-H Unimplemented marker + dominance heuristic in compute_valid_char_ratio (Bug #6)` | `crates/kebab-parse-pdf/src/text_quality.rs`, `tasks/HOTFIXES.md` |
+| 2 | Bug #7 doc-comment + SKILL.md | clap doc-comment 의 `--media` value list 에 `code` 추가 + SKILL.md line 57 동기 | `docs(cli): list 'code' in --media help string + SKILL.md (Bug #7)` | `crates/kebab-cli/src/main.rs`, `integrations/claude-code/kebab/SKILL.md` |
+| 3 | Bug #7 CLI help assertion | 신규 test file `crates/kebab-cli/tests/cli_help_smoke.rs` 의 `search_help_lists_code_in_media_values` test | `test(cli): assert 'code' in search --help output (Bug #7 regression pin)` | `crates/kebab-cli/tests/cli_help_smoke.rs` (신규) |
+| 4 | Final sanity (no commit) | workspace test + workspace clippy + optional dogfood retest | — | none |
+
+---
+
+## §2 Per-step detail
+
+### Step 1 — Bug #6 implementation
+
+#### §2.1 Files affected
+
+- `crates/kebab-parse-pdf/src/text_quality.rs` (현재 103 line — line 1-37 body, line 39-103 tests).
+- `tasks/HOTFIXES.md` (dated entry append).
+
+#### §2.2 Action
+
+**§2.2.1** — `text_quality.rs` line 1-18 (file header comment + `compute_valid_char_ratio` body) **rewrite** per spec §4.1 의 diff. 추가:
+
+- 새 const `MOJIBAKE_MARKERS: &[&str] = &["?Identity-H Unimplemented?"]` (line 8-12 위치, lopdf 0.32.0 source 추적 comment 포함).
+- `compute_valid_char_ratio()` body 의 4-단계 marker strip → trim-empty zero → dominance cap-0.3 → 기존 ratio 계산.
+- `is_valid_text_char()` (line 20-37) **변경 없음** (signature + range list 보존).
+
+**§2.2.2** — `text_quality.rs::tests` module (line 39-103) 에 2 신규 test **append**:
+
+```rust
+#[test]
+fn identity_h_marker_dominance_caps_ratio_below_threshold() {
+    let s = format!("Page 1 of 5 {}", "?Identity-H Unimplemented?".repeat(20));
+    let r = compute_valid_char_ratio(&s);
+    assert!(r <= 0.3, "marker-dominant mixed page → ratio ≤ 0.3 (OCR fallback); got {r}");
+}
+
+#[test]
+fn identity_h_marker_minority_with_long_valid_text_keeps_high_ratio() {
+    let header = "x".repeat(200);
+    let s = format!("{header} ?Identity-H Unimplemented?");
+    let r = compute_valid_char_ratio(&s);
+    assert!(r > 0.9, "marker-minority page keeps high ratio; got {r}");
+}
+```
+
+**중요 — 스펙 §4.2 wording 보정 (critic r2 NEW-1)**: spec §4.2 의 "Replace existing Bug #6 test set with two new tests" 는 stale wording. 현 `text_quality.rs::tests` 는 8 test 보유, **Identity-H marker 관련 test 0**. 즉 net change = **+2 / -0**. brief §2.1 의 "기존 test `identity_h_marker_mixed_with_some_real_text_low_ratio` 제거" 도 동일 stale — 해당 test 미존재. executor 는 8 existing test (`empty_string_zero`, `pure_ascii_one`, `pure_hangul_syllables_one`, `pure_pua_zero`, `mixed_half`, `cjk_ideograph_valid`, `hangul_jamo_valid`, `f4_fixture_ratio_under_threshold`) 모두 **보존**.
+
+**§2.2.3** — `tasks/HOTFIXES.md` 의 latest dated section 위에 entry append:
+
+```markdown
+## 2026-05-27 — Identity-H mojibake marker bypassed OCR fallback (Bug #6)
+
+- **Symptom**: `metro-korea.pdf` (Identity-H CID font without ToUnicode CMap) 의 ingest 가 `pdf_ocr_pages=0` 으로 종료. text 전체가 `?Identity-H Unimplemented?` marker 1154회 반복 (lopdf 0.32.0 emit). text-detect ratio = 1.0 → OCR fallback threshold 0.5 bypass.
+- **Root cause**: `crates/kebab-parse-pdf/src/text_quality.rs::compute_valid_char_ratio()` 의 `is_valid_text_char()` 가 ASCII printable range (0x0020..=0x007E) 를 unconditional valid 처리. marker (28 ASCII char) 는 valid 로 count.
+- **Fix**: `MOJIBAKE_MARKERS` const 도입 + marker strip after-strip 의 trim-empty → 0.0 + dominance heuristic (strip > 잔여 일 때 cap 0.3). spec ACCEPT: `docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md` §4.1. parser_version/wire schema 영향 0.
+- **User action**: 이미 `metro-korea.pdf` class 의 mojibake-heavy PDF 를 v0.20.0 pre-bugfix2 binary 로 indexed 한 경우, `kebab ingest --force-reingest <workspace>` 로 cached skip 무효화 필요 (release notes 동등 안내).
+```
+
+#### §2.3 Acceptance
+
+actionable verify command (per-step):
+
+```bash
+# A) text_quality 신규 test 2 + 기존 8 = 10 모두 green
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo test -p kebab-parse-pdf text_quality -j 4 2>&1 | tail -10
+
+# B) parse-pdf crate clean compile
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo build -p kebab-parse-pdf -j 4 2>&1 | tail -3
+
+# C) parse-pdf clippy clean (-D warnings)
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo clippy -p kebab-parse-pdf --all-targets -j 4 -- -D warnings 2>&1 | tail -5
+```
+
+기대: A 의 tail = `test result: ok. 10 passed; 0 failed`, B = `Finished`, C = warning 0.
+
+#### §2.4 Commit
+
+```bash
+git add crates/kebab-parse-pdf/src/text_quality.rs tasks/HOTFIXES.md
+git commit -m "$(cat <<'EOF'
+fix(parse-pdf): strip Identity-H Unimplemented marker + dominance heuristic in compute_valid_char_ratio (Bug #6)
+
+Why: metro-korea.pdf (Identity-H CID font without ToUnicode CMap) 의
+ingest 가 pdf_ocr_pages=0 으로 잘못 종료. lopdf 0.32.0 의 emit
+`?Identity-H Unimplemented?` marker 28 ASCII char 가 is_valid_text_char()
+의 0x0020..=0x007E range 통과 → ratio=1.0 → OCR fallback 0.5
+threshold bypass.
+
+Change: MOJIBAKE_MARKERS const + compute_valid_char_ratio() 4-단계
+(strip → trim-empty zero → dominance cap-0.3 → 기존 ratio). marker
+list extensible. is_valid_text_char() 본체 변경 0.
+
+Tests: +2 unit (dominance + minority) on top of 기존 8. parser_version
+/ wire schema 변경 0.
+
+Refs: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md
+§4.1 / §4.2 / §6 R-1.
+EOF
+)"
+```
+
+---
+
+### Step 2 — Bug #7 doc-comment + SKILL.md
+
+#### §2.5 Files affected
+
+- `crates/kebab-cli/src/main.rs` line 158-160 (실측: SearchArgs `media` 의 3-line clap doc-comment).
+- `integrations/claude-code/kebab/SKILL.md` line 57.
+
+#### §2.6 Action
+
+**§2.6.1** — `crates/kebab-cli/src/main.rs` line 158-160 의 doc-comment edit:
+
+```diff
+         /// p9-fb-36: filter by `assets.media_type` kind. Comma-separated.
+-        /// Aliases: `md` → `markdown`. Other accepted: `markdown`, `pdf`,
+-        /// `image`, `audio`, `other`. Unknown values match nothing.
+        /// Aliases: `md` → `markdown`. Other accepted: `markdown`, `pdf`,
+        /// `image`, `audio`, `code`, `other`. Unknown values match nothing.
+```
+
+(critic r2 NEW-2 보정: spec §4.3 의 1-line 표기 vs 실제 3-line clap doc-comment 차이. 실제 file 의 multi-line 분포 그대로 유지하며 line 160 의 `image`, `audio` 사이에 `code` 삽입.)
+
+**§2.6.2** — `integrations/claude-code/kebab/SKILL.md` line 57 의 edit:
+
+```diff
+-`media` (string array — IN-list of `"markdown"` | `"pdf"` | `"image"` | `"audio"` | `"other"`; alias `"md"` → `"markdown"`)
+`media` (string array — IN-list of `"markdown"` | `"pdf"` | `"image"` | `"audio"` | `"code"` | `"other"`; alias `"md"` → `"markdown"`)
+```
+
+#### §2.7 Acceptance
+
+```bash
+# A) cli crate clean compile (doc-comment edit — compile 영향 0 기대)
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo build -p kebab-cli -j 4 2>&1 | tail -3
+
+# B) SKILL.md 의 `code` substring grep
+grep -nF '"code"' integrations/claude-code/kebab/SKILL.md
+
+# C) fresh binary 의 search --help 가 `code` 노출
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo build --release -p kebab-cli -j 4 2>&1 | tail -3
+/build/out/cargo-target/target/release/kebab search --help 2>&1 | grep -F 'code'
+```
+
+기대: A = `Finished`, B = line 57 1건 hit, C = `code` 포함 1+ line.
+
+#### §2.8 Commit
+
+```bash
+git add crates/kebab-cli/src/main.rs integrations/claude-code/kebab/SKILL.md
+git commit -m "$(cat <<'EOF'
+docs(cli): list 'code' in --media help string + SKILL.md (Bug #7)
+
+Why: kebab search --media code 가 v0.18.0 부터 functional support 됨
+(MEDIA_KINDS 외 path 로 first-class 처리, schema.v1.media_breakdown.code
+존재). 그러나 SearchArgs 의 clap doc-comment + SKILL.md line 57 의
+value list 가 stale — `code` 누락. user 가 --help 만 보고 code 미지원이라
+오해 가능.
+
+Change: 2 surface 동기 — main.rs line 158-160 의 multi-line clap
+doc-comment + integrations/claude-code/kebab/SKILL.md line 57.
+Rust binary surface / wire schema 변경 0.
+
+Out of scope (follow-up): crates/kebab-mcp/tools/search.rs:44,
+crates/kebab-core/src/search.rs:32+52, crates/kebab-app/src/
+ingest_progress.rs:69, crates/kebab-cli/tests/wire_schema_breakdowns.rs:35
+도 동일 stale list 보유. spec ACCEPT (round 1c) 의 grep boundary
+밖이므로 본 round 미포함.
+
+Refs: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md
+§4.3 / §4.3a.
+EOF
+)"
+```
+
+---
+
+### Step 3 — Bug #7 CLI help assertion test
+
+#### §2.9 Files affected
+
+- `crates/kebab-cli/tests/cli_help_smoke.rs` (신규 — 기존 file list 에 미존재).
+
+#### §2.10 Action
+
+신규 file 생성. 기존 test convention (`cli_*` prefix, `Command::new(env!("CARGO_BIN_EXE_kebab"))` pattern — 참고: `cli_readonly_quiet.rs`, `cli_schema.rs`) 답습:
+
+```rust
+// crates/kebab-cli/tests/cli_help_smoke.rs
+//
+// Regression pin — `kebab search --help` 의 `--media` value list 가
+// `code` 를 노출. Bug #7 (v0.20.0 bugfix round 2 spec §4.4).
+
+#[test]
+fn search_help_lists_code_in_media_values() {
+    let out = std::process::Command::new(env!("CARGO_BIN_EXE_kebab"))
+        .args(["search", "--help"])
+        .output()
+        .expect("kebab search --help");
+    let stdout = String::from_utf8_lossy(&out.stdout);
+    assert!(
+        stdout.contains("`code`"),
+        "search --help must list 'code' as accepted --media value; stdout = {stdout}"
+    );
+}
+```
+
+#### §2.11 Acceptance
+
+```bash
+# A) 신규 test target 빌드 + 실행
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo test -p kebab-cli --test cli_help_smoke -j 4 2>&1 | tail -10
+
+# B) cli crate tests target clean compile (전체)
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo build -p kebab-cli --tests -j 4 2>&1 | tail -3
+
+# C) cli clippy clean (-D warnings) — 신규 test file 포함
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo clippy -p kebab-cli --all-targets -j 4 -- -D warnings 2>&1 | tail -5
+```
+
+기대: A = `test result: ok. 1 passed; 0 failed`, B = `Finished`, C = warning 0.
+
+#### §2.12 Commit
+
+```bash
+git add crates/kebab-cli/tests/cli_help_smoke.rs
+git commit -m "$(cat <<'EOF'
+test(cli): assert 'code' in search --help output (Bug #7 regression pin)
+
+Why: Step 2 의 doc-comment edit 가 향후 누군가 value list 를 재정렬
+하거나 alias section 으로 분리할 때 silently 사라질 risk. clap 의
+--help 렌더링 가 doc-comment 의 free-form text 라 grep-only smoke 가
+유일한 검출 수단.
+
+Change: 신규 test file (kebab-cli convention `cli_*` prefix 답습).
+CARGO_BIN_EXE_kebab 으로 fresh binary 실행, stdout 의 `code` substring
+assert. spec §4.4 의 acceptance row 1:1 mapping.
+
+Refs: docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md §4.4
+/ §5 (acceptance row 4).
+EOF
+)"
+```
+
+---
+
+### Step 4 — Final sanity (no commit)
+
+#### §2.13 Scope
+
+3 commit append 후 workspace 전수 verify + optional dogfood. **commit 발생 X** (코드 변경 0 — verification only).
+
+#### §2.14 Acceptance
+
+```bash
+# A) workspace test 전수 — 기존 1316 + 본 round +2 unit + +1 cli = 1319 expected
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo test --workspace --no-fail-fast -j 1 2>&1 | tee /tmp/v0.20-bugfix2-test.log | tail -15
+echo "exit = ${PIPESTATUS[0]:-$?}"
+
+# B) workspace clippy clean (-D warnings)
+CARGO_TARGET_DIR=/build/out/cargo-target/target cargo clippy --workspace --all-targets -j 4 -- -D warnings 2>&1 | tail -8
+
+# C) (optional) dogfood retest — metro-korea.pdf
+#    binary 의 fresh build 가 이미 Step 2 acceptance 에서 완료.
+#    --force-reingest 후 pdf_ocr_pages 가 0 → 21+ 변화 관찰.
+#    OCR latency ≈ 10 min cost — plan drafter 가 executor 에게 optional 명시.
+#    실측 corpus 가 user-private (KEBAB_WORKSPACE 또는 ~/Documents/test/) 이면 skip 가능.
+```
+
+기대: A = `test result: ok. <N> passed; 0 failed` (N ≥ 1319), B = warning 0, C = 사용자 선택 (verifier round 0 에서 평가).
+
+#### §2.15 Commit
+
+없음 (sanity-only). executor 가 sanity green 확인 후 PR push 단계로 진행.
+
+---
+
+## §3 Verifier checklist (cumulative)
+
+spec §5 의 7 row acceptance criteria 와 1:1 mapping. verifier round 0 의 actionable command:
+
+| # | Spec §5 criterion | Verifier command | Step coverage | Pass condition |
+|---|-------------------|------------------|---------------|----------------|
+| 1 | `identity_h_marker_dominance_caps_ratio_below_threshold` green | `cargo test -p kebab-parse-pdf identity_h_marker_dominance_caps_ratio_below_threshold -j 4 2>&1 \| tail -3` | Step 1 | `1 passed; 0 failed` |
+| 2 | `identity_h_marker_minority_with_long_valid_text_keeps_high_ratio` green | `cargo test -p kebab-parse-pdf identity_h_marker_minority_with_long_valid_text_keeps_high_ratio -j 4 2>&1 \| tail -3` | Step 1 | `1 passed; 0 failed` |
+| 3 | 기존 text_quality 8 test green (regression 0) | `cargo test -p kebab-parse-pdf text_quality -j 4 2>&1 \| tail -5` | Step 1 | `10 passed; 0 failed` (8 기존 + 2 신규) |
+| 4 | `search_help_lists_code_in_media_values` green | `cargo test -p kebab-cli --test cli_help_smoke -j 4 2>&1 \| tail -3` | Step 3 | `1 passed; 0 failed` |
+| 5 | SKILL.md 의 `"code"` substring 존재 | `grep -nF '"code"' integrations/claude-code/kebab/SKILL.md` | Step 2 | line 57 1 hit |
+| 6 | workspace test 전수 green | `cargo test --workspace --no-fail-fast -j 1 2>&1 \| tail -10` | Step 4 | `0 failed`, N ≥ 1319 |
+| 7 | workspace clippy clean (-D warnings) | `cargo clippy --workspace --all-targets -j 4 -- -D warnings 2>&1 \| tail -5` | Step 4 | warning 0 |
+| 8 (optional) | dogfood retest — metro-korea.pdf 의 `pdf_ocr_pages` 0 → 21+ | manual: `kebab ingest --force-reingest <ws>` 후 ingest_report.v1 의 `items[].pdf_ocr_pages` 검사 | Step 4 | `pdf_ocr_pages > 0` for metro-korea.pdf row |
+
+executor 는 row 1-7 모두 green 시 PR push gate 통과. row 8 = verifier round 0 의 optional (사용자 corpus 가용성 + 10 min cost 평가).
+
+---
+
+## §4 Risks resolution (spec §6 의 plan-level)
+
+| ID | Spec §6 status | Plan-level action |
+|----|----------------|--------------------|
+| R-1 | resolved per critic r1 (lopdf 0.32.0 = marker 1 entry) | 본 plan §2.2.1 의 source comment 가 lopdf upgrade 시 re-verify trigger. |
+| R-2 | resolved (`trim().is_empty()` cover) | Step 1 implementation 의 §2.2.1 4-단계 중 2-단계 = trim-empty zero. |
+| R-3 | resolved (wire schema 변경 0) | parser_version `"pdf-text-v1"` / chunker_version `"pdf-page-v1.1"` 보존. version cascade 영향 0 (CLAUDE.md §Versioning cascade). |
+| R-4 | resolved per critic r1 (grep boundary = `integrations/` + `crates/kebab-cli/src` + `docs/wire-schema/v1`) | Step 2 가 spec 범위 내 2 surface 모두 커버. **추가 발견 (out of scope)** → §5 OQ-4. |
+| R-5 | resolved (`bulk.rs:161` alias normalize 통해 영향 0) | 본 plan 동작 변경 0. |
+
+추가 risk — 본 plan drafter 가 식별:
+
+- **R-6 (NEW)**: Step 4 의 optional dogfood retest 가 `KEBAB_WORKSPACE` 또는 user-private corpus 의존. CI 환경에서 verify 불가 — verifier round 0 가 evidence 부재 시 row 8 skip 명시 권고.
+
+---
+
+## §5 Open questions for executor
+
+spec ACCEPT 가 명확하므로 OQ-1/2/3 모두 resolved. 본 plan drafter 가 추가 식별:
+
+- **OQ-4 (NEW)**: spec §2.2 의 R-4 grep 결과, frozen boundary 외부 surface 가 동일 stale list 보유:
+  - `crates/kebab-mcp/src/tools/search.rs:44` — MCP tool 의 `--media` doc.
+  - `crates/kebab-core/src/search.rs:32` — `MEDIA_KINDS` const = `&["markdown", "pdf", "image", "audio", "other"]`. 주의: 이 const 가 functional 일 수 있음 — `code` 는 v0.18.0 부터 separate path 로 first-class 처리 (`schema.v1.media_breakdown.code` 존재 확인 per spec §1.2). const 자체 수정은 behavior change risk 동반 → 별도 spec 으로 분리.
+  - `crates/kebab-core/src/search.rs:52` — `MediaFilter::media` doc-comment.
+  - `crates/kebab-app/src/ingest_progress.rs:69` — progress label doc-comment.
+  - `crates/kebab-cli/tests/wire_schema_breakdowns.rs:35` — test fixture array (functional, 변경 시 test 의미 영향).
+
+  **executor action**: 본 round 미포함. PR description 또는 Step 2 commit body 에 "follow-up: open issue for stale --media value list in 5 additional surfaces" 한 줄 명시 권장.
+
+- **OQ-5 (NEW)**: spec §6 의 UX consequence — pre-bugfix2 v0.20.0 user 의 `--force-reingest` 권고가 release notes 에 들어가야 하며, 별도 phase (PR review/merge 시점) 의 작업. 본 plan 의 Step 1 §2.2.3 HOTFIXES entry 가 user-facing surface 의 일부 — release notes 가 HOTFIXES 의 user action 항목을 인용 가능.
+
+---
+
+## §6 References
+
+- **Spec ACCEPT (parent contract)**: `docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md` (308 line, round 1c).
+- **Critic round 1**: `.omc/reviews/2026-05-27-v0.20-bugfix2-spec-critic-r1-result.md` (H-1 + M-1/M-2/M-3 + L-1/L-2 + NIT-1/NIT-2 + invariant audit, 9 finding 모두 spec 에 반영).
+- **Critic round 2**: `.omc/reviews/2026-05-27-v0.20-bugfix2-spec-critic-r2-result.md` (NEW-1 = §4.2 stale arithmetic, NEW-2 = §4.3 scope description drift — 본 plan §2.2.2 + §2.6.1 에 정정 반영).
+- **Plan drafter brief**: `.omc/reviews/2026-05-27-v0.20-bugfix2-plan-drafter-brief.md`.
+- **Parent design**: `docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md` §1.3 (text-detect threshold metric), §9 (version cascade).
+- **Round 1 history**: branch `feat/pdf-scanned-ocr` HEAD = `e674ff4`, 4 commit 적층 (Bug #2 source-fs, Bug #3 chunk_id collision, Bug #3 test, Bug #4 pikepdf F4 fixture).
+- **Code locations (line 실측)**:
+  - `crates/kebab-parse-pdf/src/text_quality.rs:1-103` (전체 file).
+  - `crates/kebab-cli/src/main.rs:158-160` (SearchArgs `media` clap doc-comment, 3-line multi-line attribute).
+  - `integrations/claude-code/kebab/SKILL.md:57` (search input filter 설명).
+  - `crates/kebab-cli/tests/cli_help_smoke.rs` (신규, Step 3).
+- **External source**: `lopdf-0.32.0/src/document.rs:523` (`Document::decode_text` — sole emitter of `?Identity-H Unimplemented?`).
+
+---
+
+## §7 Constraints (spec §9 + brief §9)
+
+1. **branch 변경 0** — plan 자체는 documentation only. 본 file = plan deliverable.
+2. **spec ACCEPT frozen** — round 1c body 보수 X. 본 plan 의 §2.2 / §2.6 의 wording 정정 (`Replace existing` → `+2 / -0 additive`) 은 plan 의 local note 로 명문, spec 본문 미변경.
+3. **regression 0** — workspace test N ≥ 1319.
+4. **wire schema / version cascade 변경 0** — `parser_version="pdf-text-v1"`, `chunker_version="pdf-page-v1.1"` 보존.
+5. **subagent skip** — executor 가 in-session 단일 thread 실행 (worker protocol per task assignment).
+6. **lightweight scope** — 본 plan 의 line target = 200-400 (round 1 plan = 849 line 의 1/3 미만).
+
+**Status**: DRAFT round 0 — verifier review 대기.
--- a/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix3-plan.md
+++ b/docs/superpowers/plans/2026-05-27-v0.20-sub1-bugfix3-plan.md
--- a/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md
+++ b/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md
@@ -0,0 +1,965 @@
+---
+title: v0.20.0 sub-item 1 bugfix — chunk_id collision + walker code limit + F4 fixture
+created: 2026-05-27
+status: ACCEPT (round 2 closure — Phase A complete)
+target_version: 0.20.0 (PR #189 force-update)
+parent_spec: docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md
+dogfood_evidence: .omc/reviews/2026-05-27-v0.20-dogfood-report.md
+review_history:
+  - "2026-05-27 spec round 1 critic (opus, thorough) — ACCEPT, HIGH 0 + MEDIUM 3 + LOW 2 + NIT 2"
+  - "2026-05-27 spec round 1c rewrite (opus, drafter) — MEDIUM/LOW/NIT all applied"
+  - "2026-05-27 spec round 2 closure critic (opus) — ACCEPT, 7/7 applied + 1 NIT (frontmatter status, applied here)"
+---
+
+# v0.20.0 sub-item 1 bugfix — chunk_id collision + walker code limit + F4 fixture
+
+본 spec 은 v0.20.0 sub-item 1 (scanned PDF OCR) 의 PR #189 dogfood 에서 발견된
+3 bug 의 root cause 분석 + fix design + acceptance criteria 를 명문화한다.
+후속 plan + executor 단계의 source 다.
+
+## §1 Background + dogfood evidence chain
+
+### §1.1 dogfood 환경
+
+| 항목 | 값 |
+|------|----|
+| Binary | `kebab v0.20.0` (commit `b4d9e60`) |
+| Ollama endpoint | `http://192.168.0.47:11434` (qwen2.5vl:3b) |
+| Isolated KB | `/build/cache/tmp/v0.20-dogfood/` |
+| Corpus | 9 PDF (PoC + sub-item fixture + 3 user PDF, 466 KB ~ 58 MB) |
+
+### §1.2 3 bug 의 reproducibility
+
+| Bug | Severity | Trigger | Reproducible |
+|-----|----------|---------|--------------|
+| #2 walker code limit | Important | 256 KB+ PDF/image/markdown ingest | 항상 (default config) |
+| #3 chunk_id collision | **Critical** | scanned_page2.pdf (1580 OCR chars) ingest | force-reingest 마다 |
+| #4 F4 Pages tree | Important | mojibake.pdf (F4 fixture) ingest | 항상 |
+
+### §1.3 dogfood report 인용
+
+dogfood report (`.omc/reviews/2026-05-27-v0.20-dogfood-report.md`) 의 핵심 인용:
+
+- Bug #2: `scanned=3, skipped_size_exceeded=6` — workspace 9 PDF 중 3 만 통과,
+  6 PDF (F1 466KB / F2 756KB / metro 57MB / thermal-pos 1.1MB / thermal-label
+  2.7MB / internals 820KB) walker 단계 skip.
+- Bug #3: `"DocumentStore::put_chunks (pdf): sqlite error: UNIQUE constraint
+  failed: chunks.chunk_id: ... Error code 1555: A PRIMARY KEY constraint
+  failed"` — scanned_page2.pdf chunk INSERT 단계에서 발생.
+- Bug #4: `block_count: 0, chunk_count: 0` — F4 mojibake.pdf 의 ingest 결과
+  가 PdfTextExtractor 의 "1 paragraph per page" invariant 위반.
+
+## §2 Goals + non-goals
+
+### §2.1 Goals
+
+- 3 bug 모두 v0.20.0 안 fix (PR #189 force-update path — 새 commit 들이 같은
+  branch 위에 stack).
+- parent spec
+  (`docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md`) 의 invariant 보존:
+  - §1.4 PdfTextExtractor 의 "1 Block::Paragraph per page".
+  - §3.5 post-extract OCR enrichment 의 block_id 보존 (in-place mutate path).
+  - §4.6 wire schema additive 만 (V00X migration 불필요).
+- parent plan
+  (`docs/superpowers/plans/2026-05-27-pdf-scanned-ocr-plan.md`) round 1c ACCEPT
+  의 design decisions 와 충돌 0.
+- workspace test regression 0 (`cargo test --workspace -j 1`).
+
+### §2.2 Non-goals
+
+- 새 wire schema major bump (v1 → v2) — 본 fix 들은 추가 schema 변경 0.
+- 새 V00X sqlite migration — `chunks` table DDL 변경 없음, fix 는 chunk_id 계산
+  path 한정.
+- F4 fixture 의 invariant 변경 (ToUnicode CMap 부재 + valid 1-page PDF
+  요구사항 유지).
+- 새 config knob 추가 (`[ingest.pdf].max_file_bytes` 같은 per-media-type limit
+  은 v0.21+ scope; 본 fix 는 walker 의 code path 분리만).
+
+## §3 Bug #2 — walker code limit
+
+### §3.1 Root cause (file:line evidence)
+
+`crates/kebab-source-fs/src/connector.rs:42-72` — `FsSourceConnector` 가
+`Config::new` 에서 `max_file_bytes` 와 `max_file_lines` 를
+`config.ingest.code` 단일 namespace 에서 읽는다:
+
+```rust
+Ok(Self {
+    default_root: root,
+    default_exclude: config.workspace.exclude.clone(),
+    copy_threshold_bytes,
+    skip_generated_header: config.ingest.code.skip_generated_header,
+    max_file_bytes: config.ingest.code.max_file_bytes,    // <-- code-specific
+    max_file_lines: config.ingest.code.max_file_lines,    // <-- code-specific
+})
+```
+
+`crates/kebab-source-fs/src/connector.rs:169-190` — walker 의 size check 가
+`is_oversized(...)` 호출 시 path 의 media type 무관:
+
+```rust
+if crate::code_meta::is_oversized(
+    &abs_path,
+    self.max_file_bytes,    // generic limit, applied 모든 file
+    self.max_file_lines,
+).unwrap_or(false) {
+    fs_skips.skipped_size_exceeded =
+        fs_skips.skipped_size_exceeded.saturating_add(1);
+    // ...
+    continue;
+}
+```
+
+`crates/kebab-source-fs/src/code_meta.rs:114-129` — `is_oversized(...)` 자체는
+generic helper (extension 무관):
+
+```rust
+pub(crate) fn is_oversized(path: &Path, max_bytes: u64, max_lines: u32) -> Result<bool> {
+    let meta = std::fs::metadata(path)?;
+    if meta.len() > max_bytes {
+        return Ok(true);
+    }
+    // line cap (streaming)
+    ...
+}
+```
+
+`crates/kebab-config/src/lib.rs:535-547` — `IngestCodeCfg::default()` 의
+`max_file_bytes = 262_144` (256 KB) — 대부분 PDF/image 가 이를 초과.
+
+### §3.2 Decision matrix
+
+| Option | 설명 | 장점 | 단점 |
+|--------|------|------|------|
+| **A — code path only** | walker 의 size check 를 code file (extension recognized by `code_lang_for_path`) 에만 적용 | 간단 / 기존 default behavior 보존 / Bug #2 즉시 해결 | PDF/image/markdown 의 size limit 0 — 1 GB PDF 도 walker 통과 |
+| B — per-type config | 새 `[ingest.pdf]`, `[ingest.image]`, `[ingest.markdown]` section 추가 + per-type limit | user-tunable | 새 config field × 3 + serde default + env override + tests — v0.20 hotfix scope 초과 |
+| C — generic limit + docs note | 같은 generic limit 유지하지만 의도 명문화 | code 변경 0 | UX bug 미해결 (dogfood 의 workaround config 가 production 강제) → **거부** |
+
+### §3.3 Chosen path — Option A
+
+walker 의 size cap 은 code-specific 의도. PDF/image/markdown 의 size 는
+parser 단계에서 자체 검증 (PDF 의 lopdf load_mem 은 256 KB 이상도 정상 처리,
+image 의 OCR 호출도 max_pixels 로 자체 cap). v0.21+ 에서 per-type config
+필요 시 Option B 로 진화.
+
+`is_code_file(path: &Path) -> bool` helper 추가:
+- `code_meta::code_lang_for_path(path).is_some()` = code file. 기존 helper
+  재사용으로 매핑 일관성 보장 (Tier 1 + Tier 2 basename + extension list
+  완전 동일).
+
+### §3.4 Implementation (Rust diff)
+
+`crates/kebab-source-fs/src/code_meta.rs` — `pub(crate)` helper 추가:
+
+```rust
+/// Returns true when `path`'s filename/extension is recognised as a code file
+/// (per `code_lang_for_path`). Used by the walker to apply
+/// `[ingest.code].max_file_bytes` / `max_file_lines` only to code files,
+/// not to PDF/image/markdown (which have their own size controls in their
+/// respective parsers).
+pub(crate) fn is_code_file(path: &Path) -> bool {
+    code_lang_for_path(path).is_some()
+}
+```
+
+`crates/kebab-source-fs/src/connector.rs:168-190` — walker conditional 추가:
+
+```rust
+// p10-1A-1: apply per-file generated-header + size-cap checks on files
+// that passed the override (gitignore/builtin/kebabignore) matching.
+// v0.20.0 sub-item 1 bugfix: size-cap (max_file_bytes / max_file_lines)
+// applies ONLY to code files. PDF/image/markdown bypass — their parsers
+// have their own size controls.
+if crate::code_meta::is_code_file(&abs_path)
+    && crate::code_meta::is_oversized(
+        &abs_path,
+        self.max_file_bytes,
+        self.max_file_lines,
+    )
+    .unwrap_or(false)
+{
+    fs_skips.skipped_size_exceeded =
+        fs_skips.skipped_size_exceeded.saturating_add(1);
+    push_sample(
+        &mut fs_skips.skip_examples.size_exceeded,
+        &abs_path,
+        &root,
+    );
+    tracing::debug!(
+        path = %rel_path.display(),
+        max_bytes = self.max_file_bytes,
+        max_lines = self.max_file_lines,
+        "skip: code file exceeds size cap"
+    );
+    continue;
+}
+```
+
+`skip_generated_header` 의 conditional 적용은 별개 — generated header sniff
+은 path extension 무관하게 first 512 bytes 의 ASCII content 만 본다. PDF/image
+의 binary 첫 512 byte 가 "do not edit" 같은 ASCII string 을 절대 포함하지
+않으므로 false positive 0. **`is_generated_file` 의 walker conditional 추가는
+불필요** — 기존 behavior 유지.
+
+### §3.5 Test additions
+
+`crates/kebab-source-fs/src/connector.rs` 의 기존 test module 에 추가:
+
+```rust
+#[test]
+fn size_cap_skips_only_code_files() {
+    let dir = tempfile::tempdir().unwrap();
+    let root = dir.path();
+    // 300 KB PDF (binary), 300 KB markdown (text), 300 KB Rust (code).
+    let big_blob: Vec<u8> = vec![b'x'; 300_000];
+    std::fs::write(root.join("paper.pdf"), &big_blob).unwrap();
+    std::fs::write(root.join("notes.md"), &big_blob).unwrap();
+    std::fs::write(root.join("big.rs"), &big_blob).unwrap();
+
+    let conn = FsSourceConnector::new(
+        &cfg_with_size_cap(root.to_str().unwrap(), 262_144, 5_000),
+    )
+    .unwrap();
+    let (assets, skips) = conn.scan_with_skips(&SourceScope::default()).unwrap();
+
+    let paths: Vec<_> = assets.iter().map(|a| a.workspace_path.0.clone()).collect();
+    // PDF + Markdown pass through walker.
+    assert!(paths.contains(&"paper.pdf".to_string()));
+    assert!(paths.contains(&"notes.md".to_string()));
+    // Code file gets skipped.
+    assert!(!paths.contains(&"big.rs".to_string()));
+    assert!(
+        skips.skip_examples.size_exceeded.iter().any(|p| p.contains("big.rs")),
+        "size_exceeded examples should contain only big.rs: {:?}",
+        skips.skip_examples.size_exceeded
+    );
+    assert!(
+        !skips.skip_examples.size_exceeded.iter().any(|p| p.contains("paper.pdf")),
+        "PDF must NOT appear in size_exceeded examples: {:?}",
+        skips.skip_examples.size_exceeded
+    );
+}
+```
+
+추가로 기존 test `ingest_report_counts_oversized_files_by_bytes` 의 fixture
+이름이 `huge.rs` 라서 invariant 보존됨. `ingest_report_size_cap_by_line_count`
+도 `longfile.rs` 라서 동일.
+
+## §4 Bug #3 — chunk_id collision (Critical)
+
+### §4.1 Root cause investigation
+
+#### §4.1.1 chunker 의 collision-avoidance workaround
+
+`crates/kebab-chunk/src/pdf_page_v1.rs:47-60` 의 module doc 에 collision 회피
+설명:
+
+```
+Design §4.2's chunk_id = blake3(doc_id || chunker_version || sort(block_ids)
+|| policy_hash) collides when one block (= one PDF page) is split into
+multiple chunks: every chunk on that page has identical block_ids.
+
+Workaround: feed a per-chunk variant format!("{base_policy_hash}#c{char_start}")
+into the recipe's policy_hash slot.
+```
+
+`crates/kebab-chunk/src/pdf_page_v1.rs:170-172` 의 actual call:
+
+```rust
+let per_chunk_hash = format!("{base_policy_hash}#c{char_start}");
+let chunk_id =
+    id_for_chunk(&doc.doc_id, &chunker_version, &block_ids, &per_chunk_hash);
+```
+
+여기 `char_start` = `chunk_page(...)` 의 첫 번째 tuple field = **post-overlap
+`actual_start`** (NOT 원본 segment boundary `start`).
+
+#### §4.1.2 overlap 의 actual_start 계산
+
+`crates/kebab-chunk/src/pdf_page_v1.rs:266-281`:
+
+```rust
+let actual_start = if let Some(prev) = chunks.last() {
+    let prev_min = prev.0;   // previous chunk 의 actual_start
+    let mut a = start;
+    let mut acc_o: usize = 0;
+    while a > prev_min {
+        let cl = chars[a - 1].len_utf8();
+        if acc_o + cl > overlap_bytes {
+            break;
+        }
+        acc_o += cl;
+        a -= 1;
+    }
+    a
+} else {
+    start
+};
+```
+
+`while a > prev_min` — overlap walk 는 previous chunk 의 actual_start 까지만
+back-walk. overlap_bytes 가 충분히 크고 `start - prev_min` 이 작으면
+`actual_start = prev_min`. **두 chunk 가 같은 actual_start = 같은 `#c{N}`**.
+
+#### §4.1.3 가설 검증 — F2 (1580 chars OCR)
+
+가정: F2 의 OCR text 가 첫 ~80 chars 안에 sentence-end (`.` + whitespace)
+또는 paragraph break (`\n\n`) 를 포함.
+
+- 기본 chunking policy: `target_tokens=500` → `target_bytes=1500`,
+  `overlap_tokens=80` → `overlap_bytes = min(240, 750) = 240`.
+- 한국어 char = 3 byte UTF-8. overlap_bytes=240 → 80 char 까지 back-walk.
+- 가정한 bounds = `[0, 30, ~n]` (첫 ~30 chars 안 sentence-end 1 개).
+- segment 1: start=0, chunk_end=30 → chunks.push((0, 30, ...)). `#c0`.
+- segment 2: start=30, byte_len(30, n) >> target_bytes → 단일 segment chunk.
+  - actual_start walk: a=30 → walk back while a > 0, acc_o ≤ 240.
+  - 30 chars * 3 byte = 90 byte ≤ 240. → a=0 (=prev_min) 에서 loop 종료.
+  - actual_start = 0 = prev_min.
+- chunks.push((0, n, ...)). `#c0` — **collision with chunk 1**.
+
+같은 doc 안 두 chunk 의 chunk_id input:
+- `{kind:"chunk", doc_id:doc_id_F2, chunker_version:"pdf-page-v1",
+   block_ids:[block_id_F2], policy_hash:"{base_hash}#c0"}`
+- canonical JSON 동일 → blake3 동일 → chunk_id 동일.
+
+→ `put_chunks` 의 `INSERT INTO chunks` 에서 첫 row 성공, 두 번째 row 가
+PRIMARY KEY violation.
+
+#### §4.1.4 F1 (779 chars OCR) 가 collision 안 하는 이유
+
+F1 OCR text 도 한국어이지만 character 분포가 다르거나 첫 ~80 char 안 sentence
+boundary 부재. 그 경우 bounds = `[0, n]` 또는 first boundary 가 80 char 이후
+→ chunk 2 의 actual_start 가 prev_min 이 아닌 다른 값 → distinct `#c{N}`
+값 → distinct chunk_id.
+
+→ **F2 만 collision** 이라는 dogfood 의 observation 과 일치.
+
+#### §4.1.5 dogfood report 의 가설 평가
+
+dogfood report 는 "scanned_page1 의 chunk_id 와 동일" 로 cross-doc collision
+을 추정. 본 spec 의 investigation 결과 = **intra-doc (F2 내부) collision**.
+근거:
+- chunk_id input 에 `doc_id` 포함 → 서로 다른 doc 의 chunk_id 는 자동으로 다름.
+- 같은 doc 안 두 chunk 가 같은 block_id + 같은 `#c{N}` policy_hash 면
+  identical chunk_id.
+- 가설 A (policy_hash default magic value) — 검증 안 됨, base_policy_hash 는
+  policy 의 canonical JSON blake3 (deterministic).
+- 가설 B (id_for_block 의 char_end 가 hash 의 일부) — 가능성 있지만 chunk_id
+  collision 자체와 무관 (block_id 변경은 chunk_id 변경을 produce; 다른
+  collision pattern).
+- 가설 C (chunker 의 block_ids ordering) — 가능성 있지만 single-block per
+  chunk 이므로 ordering N/A.
+- 가설 D (OCR text 가 다른 doc 와 동일 inline) — chunk_id 의 input 에 text
+  미포함, N/A.
+
+**Confirmed root cause** = 가설 C 의 variant — 단일 page 가 multi-chunk 일
+때 overlap 의 actual_start 가 prev chunk 의 actual_start 로 collapse, `#c{N}`
+suffix 동일.
+
+### §4.2 Decision matrix
+
+| Option | 설명 | 장점 | 단점 |
+|--------|------|------|------|
+| **A — segment boundary `start`** | `per_chunk_hash` 의 suffix 를 post-overlap `actual_start` 대신 segment boundary `start` 로 변경 | minimal change / segment boundary 는 monotonically increasing (chunk_page 의 seg_idx loop invariant) → 항상 distinct / chunk_id 의 semantic 의도 보존 | chunk_page 의 return tuple shape 변경 필요 |
+| B — chunk ordinal | `per_chunk_hash = "#c{ordinal}"` (page 안 chunk index 0, 1, 2, ...) | 가장 simple / segment boundary 무관 | chunk_id 의 "meaningful hash input" semantic 약화 |
+| C — (`char_start`, `char_end`) pair | `per_chunk_hash = "#c{char_start}-{char_end}"` | 두 chunk 가 같은 char_start 라도 char_end 가 다르면 distinct | char_end 도 overlap clamp 에 의해 동일 가능 (e.g. last chunk 이 두 번 분할되면) — invariant 약함 |
+| D — sequence number + char_start | `per_chunk_hash = "#c{ordinal}-{char_start}"` | invariant 완전 보장 | redundant info, hash input 가 더 길어짐 |
+
+### §4.3 Chosen path — Option A
+
+근거:
+- chunk_page 의 main loop 는 `seg_idx` 가 strictly increasing, segment
+  boundary `bounds[seg_idx]` 도 strictly increasing (bounds 가 dedup 후 unique).
+  따라서 segment boundary `start` 를 hash suffix 로 쓰면 같은 page 안 chunk
+  들의 hash input 가 보장된 distinct.
+- chunk_id 의 semantic: "어떤 segment 부터 시작한 chunk 인가" — overlap 이전
+  의 segment boundary 가 진짜 semantic origin. overlap 은 retrieval boundary
+  를 위한 enrichment.
+- chunk_page 의 return tuple 을 `(segment_start, actual_start, chunk_end,
+  slice)` 의 4-tuple 로 확장 (또는 segment_start 를 chunker loop 안에서 별도
+  track) — minimal diff.
+
+### §4.4 Implementation
+
+`crates/kebab-chunk/src/pdf_page_v1.rs` 의 `chunk_page` return signature 확장:
+
+```rust
+/// Split a single page's text into ordered chunks, each represented as
+/// `(segment_start, actual_start, chunk_end, text_slice)`.
+///
+/// - `segment_start` = pre-overlap segment boundary. Strictly increasing
+///   across the returned vec. Use this for chunk_id uniqueness suffixes.
+/// - `actual_start` = post-overlap start char index. May collapse to a
+///   previous chunk's `actual_start` under aggressive overlap policy.
+///   Use this for `SourceSpan::Page::char_start`.
+/// - `chunk_end` = chunk's end char index (exclusive).
+fn chunk_page(
+    text: &str,
+    target_bytes: usize,
+    overlap_bytes: usize,
+) -> Vec<(usize, usize, usize, String)> {
+    // ... (existing logic, but each push uses (segment_start, actual_start, chunk_end, slice))
+    chunks.push((start, actual_start, chunk_end, slice));
+    // ...
+}
+```
+
+caller 의 `chunk` method 도 동일하게 update:
+
+```rust
+for (segment_start, char_start, char_end, slice) in
+    chunk_page(&p.text, target_bytes, overlap_bytes)
+{
+    // ... existing u32 conversion + span construction ...
+    let span = SourceSpan::Page {
+        page: page_num,
+        char_start: Some(u32::try_from(char_start).expect("...")),
+        char_end: Some(u32::try_from(char_end).expect("...")),
+    };
+    let block_ids: Vec<BlockId> = vec![p.common.block_id.clone()];
+    // segment_start (pre-overlap boundary) is strictly increasing across
+    // chunks, even when overlap walk collapses actual_start to prev_min.
+    let per_chunk_hash = format!("{base_policy_hash}#c{segment_start}");
+    let chunk_id =
+        id_for_chunk(&doc.doc_id, &chunker_version, &block_ids, &per_chunk_hash);
+    // ... rest unchanged ...
+}
+```
+
+`crates/kebab-chunk/src/pdf_page_v1.rs:47-60` 의 module doc 도 동시 update —
+기존 description 의 `"#c{char_start}"` 가 새 fix 에 stale 하므로:
+
+```rust
+//! Design §4.2's chunk_id = blake3(doc_id || chunker_version || sort(block_ids)
+//! || policy_hash) collides when one block (= one PDF page) is split into
+//! multiple chunks: every chunk on that page has identical block_ids.
+//!
+//! Workaround that doesn't change the §4.2 recipe: feed a per-chunk
+//! variant `format!("{base_policy_hash}#c{segment_start}")` into the
+//! recipe's `policy_hash` slot. `segment_start` is the pre-overlap segment
+//! boundary, strictly increasing across the returned chunks even when the
+//! overlap walk collapses `actual_start` to a previous chunk's `prev_min`.
+//! Logged in tasks/HOTFIXES.md (2026-05-27 — Bug #3 second-iteration patch).
+```
+
+추가로 `tasks/HOTFIXES.md` 에 dated entry 추가 (본 fix 이 chunk_id deviation
+의 **second-iteration patch** — 첫 iteration 의 `#c{char_start}` workaround 가
+aggressive overlap case 에서 collision 을 leave 했음을 명문화):
+
+```markdown
+## 2026-05-27 — v0.20.0 sub-item 1: chunk_id `#c{char_start}` workaround
+collapses under aggressive overlap (Bug #3 second-iteration patch)
+
+**Symptom**: F2 (1580 chars OCR) ingest 시 `DocumentStore::put_chunks (pdf):
+UNIQUE constraint failed: chunks.chunk_id`. ...
+
+**Root cause**: `crates/kebab-chunk/src/pdf_page_v1.rs:170` 의 ...
+post-overlap `actual_start` 가 prev chunk 의 actual_start 로 collapse ...
+
+**Fix** (this spec, §4.4): `chunk_page` return tuple 에 `segment_start`
+추가, `per_chunk_hash` 의 suffix 를 `segment_start` 로 변경 ...
+
+**chunker_version cascade**: `pdf-page-v1` → `pdf-page-v1.1` bump
+(see §4.4.1). multi-chunk PDF page 의 chunk_id 가 변경 — design §9
+cascade trigger 로 explicit invalidation.
+
+**Amends**: spec `docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix-spec.md`
+§4.4. parent design §4.2 chunk_id recipe 자체는 unchanged (workaround layer
+의 internal computation 만 변경).
+```
+
+#### §4.4.1 chunk_id determinism 보존
+
+기존 single-chunk-per-page case (e.g. small pages, `text.len() <= target_bytes`)
+:
+- `chunk_page` 의 early return: `vec![(0, n, text.to_string())]` → 새 shape
+  로 `vec![(0, 0, n, text.to_string())]`. `segment_start = 0 = actual_start`.
+- `#c0` suffix 동일 → chunk_id 동일.
+
+multi-chunk case 의 첫 chunk:
+- segment_start = bounds[0] = 0, actual_start = start = 0 (no previous chunk).
+- `#c0` suffix 동일 → chunk_id 동일.
+
+multi-chunk case 의 second-and-later chunk:
+- 기존: `actual_start` (overlap-walked, may be == 0).
+- 새: `segment_start` = bounds[seg_idx] > 0.
+- → chunk_id 변경 (intentional, collision 회피).
+
+→ existing v0.19 (pre-OCR) PDF KB 안 multi-chunk pages 의 chunk_id 가 변경됨.
+이는 v0.20 의 force-reingest path 에서 자동 갱신.
+
+**Decision (round 1c, closes §7.2 Open Q1): chunker_version bump
+`pdf-page-v1` → `pdf-page-v1.1`** (critic round 1 M-1 권장 채택).
+
+근거:
+- 정상 multi-chunk PDF page (예: dogfood report Scenario 1 의 metro-korea.pdf
+  의 21 block / 34 chunk — Bug #3 trigger 안 한 정상 path) 의 chunk_id 가
+  internal computation 변경으로 silent 하게 다른 값으로 mapping.
+  chunker_version 을 `pdf-page-v1` 유지하면 store/embedding layer 의 cascade
+  audit 가 발생 안 함 → 사용자가 `--force-reingest` 를 명시적으로 호출하지
+  않는 한 vector store 의 chunk_id ↔ chunk_text 가 silent mismatch 가능.
+- design §9 cascade rule 의 본래 의도 = chunker algorithm 변경 시 explicit
+  version bump → store layer 의 자동 invalidation report. `pdf-page-v1.1`
+  bump 는 그 rule 의 직접 적용.
+- bump cost = zero — v0.20.0 자체가 force-update release (PR #189 단일
+  release commit 위에 cumulative bugfix stack) 이고, parent spec
+  (`2026-05-27-pdf-scanned-ocr-spec.md`) 의 OCR feature 활성화가 어차피
+  force-reingest 권장 path. single-chunk PDF page 는 chunker_version 만
+  다르면 새 doc_id chain 안에서 동일하게 cascade 재계산.
+- benefit = explicit user-facing audit trail. 다음 ingest 시 cascade
+  invalidation 이 store layer report 에 명시.
+
+cascade 의 다른 version field (parser_version / embedding_version /
+prompt_template_version / index_version) 는 unchanged — chunker layer
+한정 patch.
+
+`PdfPageV1Chunker` 의 `chunker_version()` 상수 update:
+```rust
+impl Chunker for PdfPageV1Chunker {
+    fn chunker_version(&self) -> ChunkerVersion {
+        ChunkerVersion("pdf-page-v1.1".to_string())  // was: "pdf-page-v1"
+    }
+    // ...
+}
+```
+
+`crates/kebab-chunk/src/pdf_page_v1.rs` 의 `PARSER_VERSION` 또는 const
+`CHUNKER_VERSION` 도 동시 갱신 (해당 crate 의 actual constant 명에 맞춰서).
+
+### §4.5 Test additions
+
+`crates/kebab-chunk/src/pdf_page_v1.rs` 의 `#[cfg(test)] mod tests` 에 추가:
+
+```rust
+#[test]
+fn multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids() {
+    // Regression test for v0.20.0 sub-item 1 Bug #3: post-overlap actual_start
+    // can collapse to prev_min, producing identical `#c{char_start}` suffixes
+    // and identical chunk_ids → sqlite chunks.chunk_id PRIMARY KEY violation
+    // at put_chunks INSERT time.
+    //
+    // Synthesises Korean OCR text shape: dense Korean characters (3 bytes
+    // per char) with a single early sentence-end boundary at char ~10 +
+    // long tail.
+
+    // 10 Korean chars (= 30 UTF-8 bytes) + "." + " " + ~500 more Korean chars.
+    let early_seg: String = std::iter::repeat('가').take(10).collect();
+    let tail: String = std::iter::repeat('나').take(500).collect();
+    let page_text = format!("{early_seg}. {tail}");
+
+    let doc = make_pdf_doc(&[&page_text]);
+    let policy = default_policy(500, 80);  // target=1500 byte, overlap=240 byte
+    let chunks = PdfPageV1Chunker.chunk(&doc, &policy).unwrap();
+
+    assert!(
+        chunks.len() >= 2,
+        "expected ≥2 chunks for {} byte page; got {}",
+        page_text.len(),
+        chunks.len()
+    );
+
+    // Hard invariant: all chunk_ids must be unique. Without the fix, the
+    // second chunk would have actual_start = 0 (== first chunk's
+    // actual_start) under the aggressive-overlap walk → identical `#c0`
+    // hash suffix → identical chunk_id → PRIMARY KEY violation.
+    let mut ids: Vec<&str> = chunks.iter().map(|c| c.chunk_id.0.as_str()).collect();
+    ids.sort_unstable();
+    let total = ids.len();
+    ids.dedup();
+    assert_eq!(
+        ids.len(),
+        total,
+        "all chunk_ids must be unique even when overlap walks actual_start back to prev_min"
+    );
+}
+```
+
+(round 1c L-1: 원래 round 0 의 second test
+`chunk_id_recipe_uses_segment_start_not_actual_start` 는 본 test 의
+uniqueness 검증과 redundant + 실제 assertion 이 `assert!(chunks.len() >= 2)`
+뿐이라 test name 의 의도와 mismatch — 제거.)
+
+추가로 `crates/kebab-app/tests/` 에 integration 수준의 regression test:
+
+```rust
+// crates/kebab-app/tests/multi_scanned_pdf_ingest_no_chunk_id_collision.rs (new)
+//
+// v0.20.0 sub-item 1 Bug #3 regression: 다중 scanned PDF (각자 단일 page +
+// 다른 OCR text length) 의 ingest 가 chunk_id collision 없이 모두 통과.
+//
+// Mock OCR engine (kebab-parse-image 의 MockOcrEngine 또는 inline impl) 이
+// page 마다 다른 text 길이 (예: 30 chars, 200 chars, 800 chars) return 하도록
+// 구성. real Ollama 호출 회피.
+
+#[test]
+fn multi_scanned_pdf_ingest_no_chunk_id_collision() {
+    // ... setup: 3 scanned PDF fixture, mock OCR engine, isolated KB
+    // ... assert: ingest_report.items 모두 kind != Error
+    // ... assert: store.get_chunks_count() = sum of per-PDF chunk_counts
+}
+```
+
+(round 1c NIT-1: 파일명과 함수명을 `multi_scanned_pdf_ingest_no_chunk_id_collision`
+로 통일 — 원래 round 0 의 파일명 `pdf_multi_scan_no_chunk_id_collision.rs` 는
+fn name 과 mismatch.)
+
+#### §4.5.1 Pre-condition — MockOcrEngine availability (round 1c M-3)
+
+본 integration test 는 `OcrEngine` trait 의 mock impl 을 요구. executor 단계의
+1st step:
+
+1. `grep -rn "impl OcrEngine" crates/kebab-parse-image/src/ crates/kebab-app/tests/`
+   로 MockOcrEngine 위치 확인.
+2. **현재 상태** (2026-05-27 verifier probe):
+   - `crates/kebab-parse-image/src/ocr.rs:235` — production `impl OcrEngine for OllamaVisionOcr`.
+   - `crates/kebab-app/tests/pdf_ocr_apply.rs:25` — `impl OcrEngine for MockOcrEngine` (test-only).
+3. 본 새 integration test (`multi_scanned_pdf_ingest_no_chunk_id_collision.rs`)
+   는 같은 crate (`kebab-app`) 안의 별 test binary 라 `pdf_ocr_apply.rs` 의
+   private MockOcrEngine 를 직접 import 불가. executor 의 선택지:
+   - **Option A (권장)**: `crates/kebab-app/tests/common/mock_ocr.rs` 에
+     MockOcrEngine 를 lift (per-page text 길이를 ctor argument 로 받는
+     parameterised 형태). 두 test (`pdf_ocr_apply.rs` + 본 신 test) 모두
+     `mod common;` 으로 share.
+   - **Option B**: 본 신 test 안에 inline `impl OcrEngine for LocalMock { ... }`
+     중복 정의 (test isolation 우선, share 비용 회피).
+4. 부재 시 (또는 sharing 어려움 시 — Option B 도 비현실적 시) §6 row 7 의
+   acceptance 를 **conditional downgrade** — `kebab-chunk` 의
+   unit-level invariant (§6 row 4) 만으로 Bug #3 의 core regression 핀
+   확보. integration 회피.
+
+executor 의 dependency 확인 task 의 결정 path 가 §7.2 Open Q4 에서
+closed.
+
+### §4.6 Acceptance (Bug #3 fix)
+
+- F1 (779 chars) + F2 (1580 chars) 동시 ingest 시 chunk_id collision 0.
+- `--force-reingest` 마다 collision 0.
+- 5+ scanned PDF (한국어 OCR text 100~3000 chars 분포) 의 KB 에서 collision 0.
+- `crates/kebab-chunk` 의 기존 1000-determinism test
+  (`deterministic_chunk_ids_1000`) 통과 보존.
+- workspace test regression 0.
+- new test `multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids`
+  + integration test `multi_scanned_pdf_ingest_no_chunk_id_collision` 추가.
+
+## §5 Bug #4 — F4 fixture Pages tree
+
+### §5.1 Root cause
+
+#### §5.1.1 현상
+
+```json
+{
+  "doc_path": "mojibake.pdf",
+  "kind": "new",
+  "byte_len": 22568,
+  "pdf_ocr_pages": 0,
+  "pdf_ocr_ms_total": 0,
+  "block_count": 0,
+  "chunk_count": 0,
+  "warnings": []
+}
+```
+
+PdfTextExtractor 의 invariant (§1.4 "1 Block::Paragraph per page") 위반.
+
+#### §5.1.2 lopdf get_pages() 의 reaction
+
+dogfood probe:
+- `lopdf::Document::load_mem(F4_bytes)` → OK.
+- `pdf_doc.get_pages()` → empty `BTreeMap`.
+- PDF byte stream 안 `/Type /Page` count = 1, `/Count` value = 1.
+
+→ structurally 1 page 가 존재하지만 lopdf 의 Pages tree traversal
+(`/Pages` → `/Kids` chain) 가 broken.
+
+#### §5.1.3 fixture 생성 path 분석
+
+`tests/fixtures/_synth/mojibake.py`:
+
+```python
+c = canvas.Canvas(str(dst), pagesize=A4)
+c.setFont(FONT_NAME, 12)
+y = A4[1] - 30*mm
+for line in ["Mojibake fixture (no ToUnicode CMap)", "..."]:
+    c.drawString(30*mm, y, line)
+    y -= 16
+c.save()
+
+data = dst.read_bytes()
+# pattern: "/ToUnicode <objref>" — strip indirect object reference
+new_data = re.sub(rb"/ToUnicode\s+\d+\s+\d+\s+R\b", b"", data)
+dst.write_bytes(new_data)
+```
+
+**Step 2 분석**: `re.sub` 가 `/ToUnicode N M R` byte sequence 를 제거하지만:
+- 제거된 bytes 의 length 만큼 PDF 의 byte offset 가 shift.
+- cross-reference table (`xref`) 의 offset entries 가 stale.
+- `startxref` value 의 offset 도 stale.
+
+**Step 3 의 startxref fix** (`tasks/HOTFIXES.md` 의 commit `c2cd3a7`):
+- manual byte edit `22130 → 22114` 로 startxref 갱신.
+- 그러나 xref table 자체의 individual offsets 도 stale — Pages tree 의
+  `/Kids` array 가 가리키는 indirect object 의 actual byte position 가
+  xref entry 와 mismatch.
+- lopdf 의 strict load 는 startxref + xref table 를 1차 검증; load 는 성공
+  하지만 Pages tree traversal 시 indirect object resolution fail → empty
+  Pages map.
+
+### §5.2 Fixture re-generation strategy
+
+| Option | 설명 | 장점 | 단점 |
+|--------|------|------|------|
+| **A — pikepdf** | reportlab 합성 후 pikepdf 로 open + ToUnicode 제거 + save (xref auto-regen) | proper xref regeneration / Pages tree intact / library available (pip install pikepdf) | 새 Python dependency (`pikepdf`) |
+| B — qpdf normalize | byte-edit 후 `qpdf --linearize input.pdf output.pdf` | external tool (이미 sub-item 1 acceptance criteria 에 qpdf hint 가 있음) | qpdf 의 normalize 가 broken xref 를 거부할 수 있음 (또는 ToUnicode reference 를 다시 inline 할 수 있음) |
+| C — reportlab disable ToUnicode | reportlab 의 합성 시 Type 0 font 의 ToUnicode CMap 생성 disable | byte-edit 회피 — clean | reportlab API 가 ToUnicode disable 를 직접 expose 안 함 (font 의 subclass 또는 monkeypatch 필요) |
+
+### §5.3 Chosen path — Option A (pikepdf)
+
+근거:
+- pikepdf 는 PDF 의 proper PDF surgery library — qpdf 의 Python bindings.
+- xref table 의 auto-regeneration + Pages tree 의 integrity 보존.
+- `pip install pikepdf` 로 dependency 추가 — 이미 fixture generation 용 Python
+  venv 가 reportlab 사용 중이라 추가 install 가 trivial.
+
+#### §5.3.1 ToUnicode strip 의 pikepdf approach
+
+reportlab 의 Type 0 font 에서 ToUnicode CMap reference 는 font dictionary 안
+`/ToUnicode <ref>` 로 등장. pikepdf 로 font dictionary 의 `/ToUnicode` entry 만
+제거:
+
+```python
+import pikepdf
+with pikepdf.open(str(dst), allow_overwriting_input=True) as pdf:
+    # Walk all indirect objects, delete /ToUnicode entry whenever found.
+    # PDF spec 상 /ToUnicode 는 Font dictionary 의 child 로만 등장 →
+    # false-positive 위험 practically zero. Font type 명시 check 생략 (§5.4
+    # 의 actual implementation 과 동일 형태).
+    for obj in pdf.objects:
+        if isinstance(obj, pikepdf.Dictionary):
+            if "/ToUnicode" in obj:
+                del obj["/ToUnicode"]
+    pdf.save(str(dst))
+```
+
+pikepdf 의 save 는 xref + Pages tree 의 integrity 자동 보존.
+
+### §5.4 Implementation (mojibake.py revision)
+
+`tests/fixtures/_synth/mojibake.py` 의 완전 rewrite:
+
+```python
+"""Synthesize mojibake fixture -- Type 0 font PDF without ToUnicode CMap.
+
+Strategy:
+1. reportlab 으로 Type 0 (CID) font 사용 한국어 PDF 합성 (정상 ToUnicode CMap 포함).
+2. pikepdf 로 open + font dictionary 의 /ToUnicode entry 제거 + save (xref 자동 regen).
+
+Dependency: reportlab + pikepdf. Install via `pip install reportlab pikepdf`.
+
+Usage:
+  python3 tests/fixtures/_synth/mojibake.py \
+      crates/kebab-parse-pdf/tests/fixtures/mojibake.pdf
+"""
+import sys
+from pathlib import Path
+from reportlab.lib.pagesizes import A4
+from reportlab.lib.units import mm
+from reportlab.pdfbase import pdfmetrics
+from reportlab.pdfbase.ttfonts import TTFont
+from reportlab.pdfgen import canvas
+import pikepdf
+
+DEJAVU_TTF = "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
+FONT_NAME = "DejaVuSans"
+pdfmetrics.registerFont(TTFont(FONT_NAME, DEJAVU_TTF))
+
+dst = Path(sys.argv[1])
+
+# Step 1: 정상 PDF 합성.
+c = canvas.Canvas(str(dst), pagesize=A4)
+c.setFont(FONT_NAME, 12)
+y = A4[1] - 30 * mm
+for line in [
+    "Mojibake fixture (no ToUnicode CMap)",
+    "Text extraction yields garbage \x00\x01\x02",
+]:
+    c.drawString(30 * mm, y, line)
+    y -= 16
+c.save()
+
+# Step 2: pikepdf 로 /ToUnicode reference strip + xref regeneration.
+removed = 0
+with pikepdf.open(str(dst), allow_overwriting_input=True) as pdf:
+    for obj in pdf.objects:
+        if isinstance(obj, pikepdf.Dictionary):
+            if "/ToUnicode" in obj:
+                del obj["/ToUnicode"]
+                removed += 1
+    pdf.save(str(dst))
+
+if removed == 0:
+    print("ERROR: no /ToUnicode entry found in any dictionary", file=sys.stderr)
+    sys.exit(2)
+
+# Step 3: invariant 검증 — load + page count.
+with pikepdf.open(str(dst)) as pdf:
+    n_pages = len(pdf.pages)
+    if n_pages != 1:
+        print(f"ERROR: expected 1 page, got {n_pages}", file=sys.stderr)
+        sys.exit(3)
+    # ToUnicode 부재 invariant 확인.
+    raw = Path(dst).read_bytes()
+    if b"/ToUnicode" in raw:
+        print("ERROR: /ToUnicode still present after strip", file=sys.stderr)
+        sys.exit(4)
+
+print(f"wrote {dst} ({dst.stat().st_size} bytes, ToUnicode stripped via pikepdf, 1 page)")
+```
+
+### §5.5 Test additions
+
+`crates/kebab-parse-pdf/tests/text_extractor.rs` (or relevant existing test
+file) 에 추가:
+
+```rust
+/// F4 mojibake.pdf 의 Pages tree invariant — Step 2 의 fixture re-generation
+/// (pikepdf-based) 가 lopdf 의 get_pages() 를 정상 return 하도록 보장.
+///
+/// Bug #4 regression: 이전 fixture (byte-edit + manual startxref) 는
+/// lopdf 의 strict load 는 통과시키지만 Pages tree traversal 시 broken
+/// indirect object resolution → empty pages map → block_count=0.
+#[test]
+fn mojibake_fixture_load_yields_one_page() {
+    let bytes = include_bytes!("../tests/fixtures/mojibake.pdf");
+    let doc = lopdf::Document::load_mem(bytes).expect("F4 fixture must lopdf-load");
+    let pages = doc.get_pages();
+    assert_eq!(
+        pages.len(),
+        1,
+        "F4 fixture must have exactly 1 page (Pages tree integrity)"
+    );
+}
+
+#[test]
+fn mojibake_fixture_has_no_tounicode_cmap() {
+    // Step 2 의 ToUnicode 부재 invariant.
+    let bytes = std::fs::read("tests/fixtures/mojibake.pdf").unwrap();
+    let count = bytes.windows(b"/ToUnicode".len())
+        .filter(|w| *w == b"/ToUnicode")
+        .count();
+    assert_eq!(count, 0, "F4 fixture must not contain /ToUnicode marker");
+}
+
+#[test]
+fn pdf_text_extractor_on_mojibake_yields_one_block() {
+    // PdfTextExtractor 의 invariant: 1 Block::Paragraph per page.
+    // F4 fixture 의 ToUnicode 부재 → text extraction yields garbage 또는
+    // empty → 1 empty Block::Paragraph + "scanned candidate" warning.
+    let bytes = include_bytes!("../tests/fixtures/mojibake.pdf");
+    // ... ExtractContext setup + extractor.extract(&ctx, bytes) ...
+    let canonical = extractor.extract(&ctx, bytes).unwrap();
+    assert_eq!(canonical.blocks.len(), 1, "expected 1 Block::Paragraph per page");
+    // text 는 garbage 또는 empty — invariant 는 block 자체의 존재.
+    let warning_present = canonical.provenance.events.iter().any(|e| {
+        matches!(e.kind, ProvenanceKind::Warning)
+            && e.note.as_ref().is_some_and(|n| n.contains("scanned candidate"))
+    });
+    assert!(warning_present || !canonical.blocks[0].text_is_empty(),
+            "text-detect first 의 empty fallback 시 scanned-candidate warning 필수");
+}
+```
+
+### §5.6 Acceptance (Bug #4 fix)
+
+- F4 fixture re-generation 후 `lopdf::Document::load_mem(...).get_pages().len() = 1`.
+- F4 fixture 의 ToUnicode CMap 부재 invariant 보존
+  (`grep -c "/ToUnicode" mojibake.pdf` = 0).
+- PdfTextExtractor 의 F4 ingest 시 `block_count = 1`,
+  warning `"page1 empty (scanned candidate)"` 또는 garbage text.
+- dogfood retest 시 mojibake.pdf 의 `block_count: 1`,
+  `chunk_count: 0~1` (depending on text content).
+- 기존 `text_extractor_regression.rs` 의 F4 baseline 갱신 — old baseline 자체
+  가 broken invariant 의 snapshot 이라 update 필요.
+- workspace test regression 0.
+
+## §6 Acceptance criteria (consolidated)
+
+| # | Verifier | Bug | 명령 |
+|---|---------|-----|------|
+| 1 | walker bypasses size cap for PDF | #2 | `cargo test -p kebab-source-fs size_cap_skips_only_code_files` |
+| 2 | walker still skips oversized code files | #2 | `cargo test -p kebab-source-fs ingest_report_counts_oversized_files_by_bytes` |
+| 3 | 256KB+ PDF/markdown ingest default config | #2 | dogfood retest: `kebab ingest` 시 `skipped_size_exceeded = 0` for non-code |
+| 4 | chunker collision regression test | #3 | `cargo test -p kebab-chunk multi_chunk_page_with_aggressive_overlap_produces_unique_chunk_ids` |
+| 5 | chunker determinism preserved | #3 | `cargo test -p kebab-chunk deterministic_chunk_ids_1000` |
+| 6 | chunker overlap clamp preserved | #3 | `cargo test -p kebab-chunk overlap_clamped_when_overlap_exceeds_target` |
+| 7 | integration: multi-scanned PDF ingest (conditional — §4.5.1 의 MockOcrEngine share 가능 시) | #3 | `cargo test -p kebab-app multi_scanned_pdf_ingest_no_chunk_id_collision` |
+| 8 | dogfood: F1 + F2 force-reingest | #3 | dogfood retest: `kebab ingest --force-reingest` 시 errors = 0 (encrypted 제외) |
+| 9 | F4 fixture lopdf 1-page invariant | #4 | `cargo test -p kebab-parse-pdf mojibake_fixture_load_yields_one_page` |
+| 10 | F4 fixture ToUnicode 부재 invariant | #4 | `cargo test -p kebab-parse-pdf mojibake_fixture_has_no_tounicode_cmap` |
+| 11 | F4 PdfTextExtractor 1-block invariant | #4 | `cargo test -p kebab-parse-pdf pdf_text_extractor_on_mojibake_yields_one_block` |
+| 12 | dogfood: F4 ingest yields block_count=1 | #4 | dogfood retest: mojibake.pdf 의 ingest item `block_count: 1` |
+| 13 | workspace clippy clean | all | `cargo clippy --workspace --all-targets -- -D warnings` |
+| 14 | workspace full test pass | all | `cargo test --workspace --no-fail-fast -j 1` |
+| 15 | dogfood end-to-end 9 PDF | all | dogfood retest: 9 PDF 모두 ingest, errors = 2 (encrypted only) |
+| 16 | chunker_version cascade — final value | #3 | `grep -nE 'pdf-page-v[0-9.]+' crates/kebab-chunk/src/pdf_page_v1.rs` 결과가 `"pdf-page-v1.1"` (round 1c M-1 결정) |
+
+## §7 Risks + open questions
+
+### §7.1 Risks
+
+- **Bug #3 fix 가 chunk_id 변경**: multi-chunk PDF page (pre-OCR 시점에 1500
+  byte 초과 page) 의 chunk_id 가 변경됨. 사용자가 `--force-reingest` 1회
+  필요. v0.20.0 force-update path 라 acceptable (user 가 어차피 fresh
+  ingest). README 또는 release note 에 명시.
+- **Bug #2 fix 의 side-effect**: 1 GB 이상의 PDF 가 walker 통과 → lopdf 의
+  load_mem 가 메모리 폭발 위험. v0.20 scope 외 (Phase 9 부터 streaming
+  parser 검토 — design §9.2 의 future scope). 본 fix 에서는 acceptable.
+- **Bug #4 fix 의 fixture binary 변경**: F4 mojibake.pdf 의 SHA256 가 변경
+  → git LFS / binary diff 의 noise. `text_extractor_regression.rs` 의
+  baseline 도 새 fixture 의 output 으로 update — 한 commit 안 동시 처리.
+- **pikepdf install requirement**: fixture re-generation 시 `pip install
+  pikepdf` 필요. CI 환경 (만약 fixture regeneration 이 CI 의 일부) 의
+  Python dependency 추가 — 본 spec 의 fix 는 fixture 자체를 commit 하므로
+  generation 은 1회성, CI 의존성 미발생.
+
+### §7.2 Open questions
+
+1. **chunker_version bump 의 cost-benefit**: ✅ **CLOSED (round 1c M-1)** —
+   `pdf-page-v1` → `pdf-page-v1.1` bump 결정. cascade audit trail explicit
+   + v0.20 force-update path 라 cost zero. detail = §4.4.1 의 "Decision
+   (round 1c, closes §7.2 Open Q1)" 단락.
+2. **Bug #2 의 Option B (per-type config) 의 v0.20 scope inclusion**: 본 spec
+   은 v0.21+ 로 defer. critic round 1 ACCEPT — v0.20 안 inclusion 권고 없음.
+3. **F4 fixture 의 invariant**: critic round 1 ACCEPT — ToUnicode 부재 +
+   valid Pages tree 조합은 pikepdf 의 proper PDF surgery 로 정확히 reproducible.
+   Step 2 의 design (mojibake.py rewrite) sound.
+4. **integration test 의 mock OCR**: ✅ **CLOSED (round 1c M-3)** —
+   `crates/kebab-app/tests/pdf_ocr_apply.rs:25` 에 이미 `impl OcrEngine for
+   MockOcrEngine` 존재. executor 의 share path (Option A — `tests/common/
+   mock_ocr.rs` lift) 또는 inline 중복 (Option B) 결정만 남음. share 가 불가
+   능 시 §6 row 7 의 conditional downgrade — detail = §4.5.1 의 "Pre-condition
+   — MockOcrEngine availability" 단락.
+5. **chunk_page tuple shape 변경**: Option A 의 4-tuple `(segment_start,
+   actual_start, chunk_end, slice)` 가 외부 callers 에 영향을 주는가?
+   `chunk_page` 는 module-private (`fn chunk_page`) 이라 외부 caller 0,
+   안전. critic round 1 ACCEPT.
+
+## §8 References
+
+- dogfood report: `.omc/reviews/2026-05-27-v0.20-dogfood-report.md`
+- parent spec (frozen): `docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md`
+- parent plan (round 1c ACCEPT): `docs/superpowers/plans/2026-05-27-pdf-scanned-ocr-plan.md`
+- source code (root cause evidence):
+  - `crates/kebab-source-fs/src/connector.rs` (Bug #2)
+  - `crates/kebab-source-fs/src/code_meta.rs` (is_oversized + code_lang_for_path)
+  - `crates/kebab-config/src/lib.rs` (IngestCodeCfg)
+  - `crates/kebab-core/src/ids.rs` (id_for_chunk / id_for_block recipes)
+  - `crates/kebab-chunk/src/pdf_page_v1.rs` (PdfPageV1Chunker + chunk_page)
+  - `crates/kebab-app/src/pdf_ocr_apply.rs` (post-extract OCR enrichment)
+  - `crates/kebab-app/src/lib.rs:1769-1968` (ingest_one_pdf_asset wiring)
+  - `crates/kebab-store-sqlite/src/documents.rs:103-155` (put_chunks DELETE+INSERT)
+  - `migrations/V001__init.sql:80-94` (chunks table DDL — chunk_id PRIMARY KEY)
+  - `tests/fixtures/_synth/mojibake.py` (Bug #4 fixture source)
+- design §3.4 (SourceSpan::Page), §3.5 (Chunk + chunk_id recipe),
+  §4.2 (id_from canonical JSON), §5.2 (walker builtin blacklist),
+  §9 (versioning cascade).
--- a/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md
+++ b/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix2-spec.md
@@ -0,0 +1,308 @@
+---
+title: "v0.20.0 sub-item 1 bugfix round 2 — Identity-H mojibake marker + CLI --media help text"
+created: 2026-05-27
+status: "DRAFT round 1c"
+parent_spec: docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md
+contract_sections: ["§1.3 (text-detect threshold metric)", "§9 (version cascade)"]
+related_specs:
+  - docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md
+  - docs/superpowers/specs/2026-04-27-kebab-final-form-design.md
+related_dogfood:
+  - .omc/reviews/2026-05-27-v0.20-bugfix-dogfood-report.md (Bug #6 + #7)
+---
+
+# v0.20.0 sub-item 1 bugfix round 2 — Identity-H mojibake marker + CLI --media help text
+
+## §1 Problem statement
+
+### §1.1 Bug #6: Identity-H Unimplemented marker bypasses mojibake detection
+
+**Symptom**: `metro-korea.pdf` (58 MB, Identity-H CID font without ToUnicode CMap) ingests with `pdf_ocr_pages=0`. Full text contains `?Identity-H Unimplemented?` marker 1154 times. All 21 pages + 34 chunks are indexable, but content is unusable garbage — repeated marker literal instead of readable text.
+
+**Root cause**: `crates/kebab-parse-pdf/src/text_quality.rs` lines 9-37. The function `compute_valid_char_ratio()` via `is_valid_text_char()` treats ASCII printable range `0x0020..=0x007E` as unconditionally valid. lopdf emits `?Identity-H Unimplemented?` (28 ASCII printable chars) when it cannot decode a CID font lacking ToUnicode CMap. Result: valid_ratio = 1.0 → exceeds OCR fallback threshold 0.5 → text-detect first-pass incorrectly classifies mojibake as valid text → `pdf_ocr_pages` stays 0, no OCR fallback triggered.
+
+**Design intent deviation**: Parent spec §1.3 (line 74) explicitly states "ratio metric judges mojibake page as scanned candidate." PoC example "֥ᬵᯝ₞e ࠦᯱᖝ░" (custom font, no ToUnicode) should trigger OCR. **Implementation gap**: literal ASCII marker case (Identity-H font) was not anticipated.
+
+### §1.2 Bug #7: CLI `--media` help text omits `code` from valid values
+
+**Symptom**: `kebab search --help` lists `--media` accepted values as "markdown, pdf, image, audio, other" — `code` is missing.
+
+**Actual behavior**: `kebab search "main" --media code --json -k 5` returns 5 hits (code/script.sh, code/rust_sample/src/main.rs, etc.). Schema `media_breakdown` includes `code: 6` as first-class. Functional correctness is complete; **CLI doc-comment is outdated only**.
+
+**Root cause**: `crates/kebab-cli/src/main.rs:148-165`. SearchArgs `--media` field clap doc-comment omits `code`. clap's `--help` renderer quotes this doc-comment directly.
+
+---
+
+## §2 Scope + non-scope
+
+### §2.1 Included in this spec
+
+- **Bug #6 fix**: Add known mojibake marker stripping to `compute_valid_char_ratio()`.
+- **Bug #6 test**: Three new unit tests covering Identity-H / Identity-V markers (full-text, mixed-text cases).
+- **Bug #6 regression**: Verify existing 8 text_quality unit tests remain green.
+- **Bug #7 fix**: Update CLI `--media` doc-comment to include `code`.
+- **Bug #7 test**: Assert that `kebab search --help` output contains "`code`" substring.
+- **Traceability**: Link both fixes to parent spec §1.3 design intent.
+
+### §2.2 Explicitly out of scope
+
+**Bug #8 candidate (falsified)**: V007 trigram tokenizer already applied; 2-character query limitation is design-level constraint, not a bug. Handled in prior dogfood report §Bug #8.
+
+**Non-bug observations**:
+- `--readonly + ingest` exit=0: Graceful refusal per CLAUDE.md intent (exit codes 0/1/2/3 unchanged; `error.v1.code` handles agent branching).
+- Ask phrasing-sensitive refusal: RAG corner case; not a code defect.
+- Binary staleness: Environmental artifact, not applicable to spec.
+
+**Ancillary risks**:
+- Scan for other `--media` doc-comment locations (R-4): Plan drafter to use grep; not blockers for this spec.
+- Other lopdf unimplemented markers (R-1): Plan drafter to inspect lopdf source; marker array is extensible.
+
+---
+
+## §3 Decisions
+
+### §3.1 Bug #6: Known mojibake marker stripping
+
+Strip known mojibake marker substrings **before ratio calculation**, then force ratio to 0.0 if remainder is empty after marker removal. When stripped characters exceed remaining characters (marker dominance), cap ratio at 0.3 to trigger OCR fallback on marker-heavy mixed pages.
+
+**Rationale**: lopdf's unimplemented CID font handling consistently emits specific ASCII marker strings. Hardcoding them is lightweight, deterministic, and covers the known failure mode without requiring expensive heuristics (e.g., ML-based gibberish detection). Pages like `metro-korea.pdf` may contain mostly mojibake body text with small valid headers; the marker-dominance check ensures such pages fall below the 0.5 OCR threshold.
+
+**Marker list**: `?Identity-H Unimplemented?` only. lopdf 0.32.0 emits exactly one marker (verified per critic round 1 probe). Extensible if future lopdf versions emit additional markers.
+
+### §3.2 Bug #7: CLI doc-comment update
+
+Add `code` to the comma-separated list of valid `--media` values in the SearchArgs field's clap doc-comment. Single-line edit; no functional or schema changes.
+
+### §3.3 Parent spec traceability
+
+Both fixes uphold parent spec §1.3:
+- Bug #6 ensures mojibake pages (Text CMap-missing fonts) trigger OCR fallback per design intent.
+- Bug #7 corrects CLI documentation to match actual schema (first-class `code` media type supported since v0.18.0).
+
+No changes to parser_version, chunker_version, or wire schema.
+
+---
+
+## §4 Implementation specification
+
+### §4.1 Bug #6: text_quality.rs diff
+
+**File**: `crates/kebab-parse-pdf/src/text_quality.rs`
+
+**Change**:
+1. Add constant array of known mojibake markers (lines 8–10):
+   ```rust
+   // Source of truth: lopdf-0.32.0/src/document.rs:523 (Document::decode_text).
+   // Only one Unimplemented marker is emitted by lopdf 0.32.0; other CMap
+   // encodings fall through to `String::from_utf8_lossy(bytes)`, which yields
+   // PUA / replacement-char territory already covered by `pure_pua_zero`.
+   // Re-verify on lopdf dependency upgrade.
+   const MOJIBAKE_MARKERS: &[&str] = &[
+       "?Identity-H Unimplemented?",
+   ];
+   ```
+
+2. Refactor `compute_valid_char_ratio()` (lines 39–106):
+   ```rust
+   pub fn compute_valid_char_ratio(s: &str) -> f32 {
+       // 1) Strip known mojibake markers before counting valid chars.
+       //    Identity-H CID fonts without ToUnicode CMap emit ASCII-only marker
+       //    substrings (bypassing PUA detection).
+       let mut cleaned: String = s.to_string();
+       // `had_marker` guard preserves prior behavior for whitespace-only input
+       // (returns ratio of whitespace validity, not 0.0) when no markers found.
+       // With markers stripped, the guard enables the trim-empty check.
+       let mut had_marker = false;
+       for marker in MOJIBAKE_MARKERS {
+           if cleaned.contains(marker) {
+               had_marker = true;
+               cleaned = cleaned.replace(marker, "");
+           }
+       }
+       // 2) Whitespace-only cleaned text → 0.0 (marker-only page).
+       if had_marker && cleaned.trim().is_empty() {
+           return 0.0;
+       }
+       // 3) Marker-dominance heuristic — when stripped chars exceed remaining
+       //    chars (i.e. marker > 50% of original), the page is "mostly mojibake
+       //    with some decodeable page-furniture" (e.g. metro-korea.pdf has
+       //    header text in a separate font + body that is Identity-H CID).
+       //    Force ratio downward to trigger OCR fallback (parent spec §1.3 intent).
+       if had_marker {
+           let stripped_chars = s.len().saturating_sub(cleaned.len());
+           if stripped_chars > cleaned.len() {
+               // Marker dominates — cap ratio at 0.3 (below 0.5 OCR threshold).
+               // The 0.3 cap (not 0.0) preserves a small signal that some text
+               // WAS decodeable, useful for downstream metrics if ever exposed.
+               let mut total = 0u32;
+               let mut valid = 0u32;
+               for c in cleaned.chars() {
+                   total += 1;
+                   if is_valid_text_char(c) {
+                       valid += 1;
+                   }
+               }
+               let raw_ratio = if total == 0 { 0.0 } else { valid as f32 / total as f32 };
+               return raw_ratio.min(0.3);
+           }
+       }
+       // 4) Otherwise compute ratio on cleaned text (existing logic).
+       let mut total = 0u32;
+       let mut valid = 0u32;
+       for c in cleaned.chars() {
+           total += 1;
+           if is_valid_text_char(c) {
+               valid += 1;
+           }
+       }
+       if total == 0 {
+           return 0.0;
+       }
+       valid as f32 / total as f32
+   }
+   ```
+
+**Invariants preserved**:
+- Function signature and return type unchanged (→ byte-identical caller surface).
+- Existing character category logic (hangul, CJK, Latin-1) unmodified.
+- Empty-string behavior (return 0.0) preserved.
+
+### §4.2 Bug #6: Unit tests
+
+Replace existing Bug #6 test set with two new tests reflecting marker-dominance heuristic:
+
+```rust
+#[test]
+fn identity_h_marker_dominance_caps_ratio_below_threshold() {
+    // metro-korea.pdf-class: 20× marker (560 char) + 11 char ASCII header.
+    // Without dominance heuristic: ratio = 11/11 = 1.0 (bypasses OCR).
+    // With dominance heuristic: ratio ≤ 0.3 (triggers OCR fallback).
+    let s = format!("Page 1 of 5 {}", "?Identity-H Unimplemented?".repeat(20));
+    let r = compute_valid_char_ratio(&s);
+    assert!(r <= 0.3, "marker-dominant mixed page → ratio ≤ 0.3 (OCR fallback); got {r}");
+}
+
+#[test]
+fn identity_h_marker_minority_with_long_valid_text_keeps_high_ratio() {
+    // Inverse case: short marker noise + long valid text → ratio stays high
+    // (no false OCR trigger on otherwise-good pages).
+    let header = "x".repeat(200);  // 200 char valid ASCII
+    let s = format!("{header} ?Identity-H Unimplemented?");  // 1× marker = 26 char
+    let r = compute_valid_char_ratio(&s);
+    assert!(r > 0.9, "marker-minority page keeps high ratio; got {r}");
+}
+```
+
+**Regression preservation**: Existing 8 tests (`empty_string_zero`, `pure_ascii_one`, `pure_hangul_syllables_one`, `pure_pua_zero`, `mixed_half`, `cjk_ideograph_valid`, `hangul_jamo_valid`, `f4_fixture_ratio_under_threshold`) must all remain green.
+
+### §4.3 Bug #7: CLI doc-comment diff
+
+**File**: `crates/kebab-cli/src/main.rs` (SearchArgs field, lines ~150–160)
+
+**Change**:
+```diff
+-/// p9-fb-36: filter by `assets.media_type` kind. Comma-separated. Aliases: `md` → `markdown`. Other accepted: `markdown`, `pdf`, `image`, `audio`, `other`. Unknown values match nothing
+/// p9-fb-36: filter by `assets.media_type` kind. Comma-separated. Aliases: `md` → `markdown`. Other accepted: `markdown`, `pdf`, `image`, `audio`, `code`, `other`. Unknown values match nothing
+```
+
+### §4.3a Bug #7 integration: `integrations/claude-code/kebab/SKILL.md:57` simultaneous update
+
+Per CLAUDE.md §Wire schema v1 invariant — in-tree integration docs must be synchronized when wire surface changes. This round has no wire schema change, but SKILL.md line 57 exhibits the same regression as §4.3 (Bug #7):
+
+**File**: `integrations/claude-code/kebab/SKILL.md` (line 57)
+
+**Change**:
+```diff
+-`media` (string array — IN-list of `"markdown"` | `"pdf"` | `"image"` | `"audio"` | `"other"`; alias `"md"` → `"markdown"`)
+`media` (string array — IN-list of `"markdown"` | `"pdf"` | `"image"` | `"audio"` | `"code"` | `"other"`; alias `"md"` → `"markdown"`)
+```
+
+### §4.4 Bug #7: CLI help assertion
+
+Add test to `crates/kebab-cli/tests/` (or extend existing help snapshot test):
+
+```rust
+#[test]
+fn search_help_lists_code_in_media_values() {
+    let out = std::process::Command::new(env!("CARGO_BIN_EXE_kebab"))
+        .args(["search", "--help"])
+        .output()
+        .expect("kebab search --help");
+    let stdout = String::from_utf8_lossy(&out.stdout);
+    assert!(stdout.contains("`code`"), "search --help must list 'code' as accepted --media value");
+}
+```
+
+### §4.5 Version cascade impact (CLAUDE.md §Versioning cascade)
+
+- **parser_version**: `"pdf-text-v1"` — unchanged. Text-detect threshold is internal metric, not surface.
+- **chunker_version**: `"pdf-page-v1.1"` — unchanged (no chunker logic affected).
+- **wire schema**: No new fields, no schema version bump. `compute_valid_char_ratio()` is internal to `PdfTextExtractor::extract()`.
+
+---
+
+## §5 Acceptance criteria
+
+- [ ] Text_quality unit test: `identity_h_marker_dominance_caps_ratio_below_threshold` passes.
+- [ ] Text_quality unit test: `identity_h_marker_minority_with_long_valid_text_keeps_high_ratio` passes.
+- [ ] Regression: All 8 existing text_quality tests remain green (no ratio behavior changes for valid text).
+- [ ] CLI help assertion: `cargo test search_help_lists_code_in_media_values` passes.
+- [ ] SKILL.md integration: `grep -F '"code"' integrations/claude-code/kebab/SKILL.md` returns ≥1 line.
+- [ ] Full workspace test suite: `cargo test --workspace --no-fail-fast -j 1` green (clippy + unit + integration).
+- [ ] Fresh binary builds: `CARGO_TARGET_DIR=/build/out/cargo-target/target cargo build --release -p kebab-cli` succeeds.
+
+---
+
+## §6 Risks + open questions
+
+### Identified risks
+
+**R-1 — Other lopdf unimplemented markers** (resolved per critic round 1 probe): lopdf 0.32.0 emits exactly one marker — `?Identity-H Unimplemented?` at `lopdf-0.32.0/src/document.rs:523` (`Document::decode_text`). Other CMap encoding arms (`UniCNS`, `UniJIS`, `UniKS`, `GBK-EUC`, `Adobe-*`) fall through to `String::from_utf8_lossy(bytes)` → PUA / replacement-char territory (already covered by `pure_pua_zero` test). Marker array adequacy = OK for current lopdf pin. **Re-verify on lopdf dependency upgrade.**
+
+**R-2 — Whitespace-only edge case after stripping**: Handled by `.trim().is_empty()` check; returns 0.0 as intended.
+
+**R-3 — Version/wire schema impact**: None. text_quality is internal threshold metric, not exposed to wire schema or version cascade.
+
+**R-4 — Other `--media` help locations** (revised per critic): `--media` value list is scattered across 3 surfaces — `crates/kebab-cli/src/main.rs:157–159` (CLI doc-comment, covered by §4.3), `integrations/claude-code/kebab/SKILL.md:57` (skill doc, covered by §4.3a), `crates/kebab-cli/tests/cli_help_smoke.rs` (test, covered by §4.4). Plan drafter to run `grep '\bmedia\b' integrations/ crates/kebab-cli/src docs/wire-schema/v1` to confirm no additional surfaces exist.
+
+**R-5 — Bulk mode media field parsing**: `crates/kebab-app/src/bulk.rs:161` handles media field parsing independently; string doc-comment update does not affect functional correctness.
+
+### Open questions
+
+**OQ-1 — Marker case sensitivity**: Does lopdf always emit markers in exact case `?Identity-H Unimplemented?`? Verify with lopdf source. If case variations exist, use case-insensitive matching or extend array.
+
+**OQ-2 — Marker stripping threshold policy** (resolved via §4.1 marker-dominance heuristic): When stripped characters exceed remaining characters, ratio is capped at 0.3 to trigger OCR fallback. This ensures marker-dominant mixed pages (e.g., 99% marker + 1% valid header) do not bypass OCR despite the header's high ratio. Design intent (parent spec §1.3) is upheld: all mojibake pages trigger OCR fallback.
+
+**OQ-3 — Alias expansion scope**: Bug #7 explicitly omits new aliases (e.g., `src` → `code`). Single additive fix to doc-comment, no enum variant changes.
+
+### UX consequence — pre-bugfix2 v0.20 user's `--force` re-ingest
+
+This round preserves version cascade (no `parser_version` bump). The `try_skip_unchanged` path will match files indexed pre-bugfix2 with same `parser_version="pdf-text-v1"` + hash. Pre-indexed `metro-korea.pdf`-class pages will NOT automatically re-route through the corrected text-detect → OCR fallback.
+
+**User action required**: Explicit `kebab ingest --force-reingest <workspace>` to purge cached skip decisions and re-process affected files.
+
+**Release notes** (v0.20.1 or whichever version ships this bugfix) **MUST include**: "If you indexed mojibake-heavy PDFs (esp. metro-korea.pdf class) on v0.20.0 pre-bugfix2, run `kebab ingest --force-reingest <workspace>` to apply the improved text detection. Otherwise, `ingest` will skip unchanged files and OCR fallback will not trigger." + link to design §9 cascade explanation.
+
+**Documentation updates** (same PR as code): README + HANDOFF + ARCHITECTURE per `feedback_readme_sync_rule` memory — mention the `--force-reingest` step in release highlights or migration notes.
+
+Deliberate design: automatic migration risks wedging stable v0.20.0 KBs. Manual `--force-reingest` is the correct escape hatch (parent spec §1.7 line 126–128 precedent).
+
+---
+
+## §7 References
+
+- **Parent spec**: docs/superpowers/specs/2026-05-27-pdf-scanned-ocr-spec.md §1.3 (line 74), §1.4, §9
+- **Dogfood evidence**: .omc/reviews/2026-05-27-v0.20-bugfix-dogfood-report.md §Bug #6, §Bug #7
+- **Critic result**: .omc/reviews/2026-05-27-v0.20-bugfix2-spec-critic-r1-result.md (findings H-1 through NIT-2, parent invariant audit)
+- **External source**: lopdf-0.32.0/src/document.rs:523 (`Document::decode_text` — sole emitter of `?Identity-H Unimplemented?` marker)
+- **Code locations**:
+  - text_quality.rs: `crates/kebab-parse-pdf/src/text_quality.rs:9-106`
+  - CLI help: `crates/kebab-cli/src/main.rs:157–159`
+  - Skill integration: `integrations/claude-code/kebab/SKILL.md:57`
+  - CLI test: `crates/kebab-cli/tests/` (search_help_lists_code_in_media_values)
+
+---
+
+**Status**: Round 1c rewrite COMPLETE. All 9 critic findings (H-1 + M-1/M-2/M-3 + L-1/L-2 + NIT-1/NIT-2 + invariant audit) applied in-session.
+
+**Prior round reference**: Round 1 commits (d9acda5, 436fd01, 241ded5, e674ff4) are merged on branch; this round is independent (text_quality.rs vs. source-fs/connector.rs + chunk/pdf_page_v1.rs).
--- a/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix3-spec.md
+++ b/docs/superpowers/specs/2026-05-27-v0.20-sub1-bugfix3-spec.md
@@ -0,0 +1,410 @@
+---
+title: "v0.20.0 sub-item 1 bugfix round 3 — final-dogfood findings"
+created: 2026-05-27
+status: DRAFT
+round: 1c
+parent_spec: 2026-04-27-kebab-final-form-design.md
+contract_sections:
+  - "1.1 (ask streaming)"
+  - "2.2 (error handling)"
+  - "2.4 (JSON wire schema)"
+  - "3.1 (config XDG)"
+  - "4.1 (capabilities schema)"
+source_report: .omc/reviews/2026-05-27-v0.20-final-dogfood-report.md
+---
+
+# v0.20.0 sub-item 1 bugfix round 3 — final-dogfood findings
+
+Post-bugfix2 final dogfood (2026-05-27) 에서 발견된 **5개 bug** 의 fix design. PR #189 force-update (base=main). Spec scope: root cause + fix decision + acceptance criteria + parent spec traceability. Bug #12 falsified (scope 외). Fix 5개 모두 trivial ~ small refactor (기존 1350 test + 추가 5+ test).
+
+---
+
+## §1 Problem statement
+
+### §1.1 Bug #9: capabilities false negative (Critical)
+
+`kebab schema --json` 의 `capabilities.streaming_ask` 와 `capabilities.single_file_ingest` 가 모두 `false` hardcoded. 그러나 실제 구현:
+- `kebab ask --stream` → `answer_event.v1` ndjson events 정상 emit (191 events 검증).
+- `kebab ingest-file <path>` → `ingest_report.v1` 신규/갱신 정상.
+- `kebab ingest-stdin --title <T>` → 정상.
+
+**Impact**: MCP host, Claude Code skill 등 agent 가 `capabilities: { streaming_ask: false, single_file_ingest: false }` 보고 routing 결정 시 false negative. user 가 실제 동작하는 feature 를 사용 불가능하다고 오인.
+
+### §1.2 Bug #10: config fail-fast (UX)
+
+```bash
+kebab search "rust" --config /tmp/nonexistent.toml --json
+# exit=0, {"hits":[],"schema_version":"search_response.v1"}
+```
+
+explicit path 가 missing 시 silent fallback to default config (XDG path). debugging nightmare — typo 또는 wrong path 가 0 hit 으로만 surface.
+
+### §1.3 Bug #11: OCR timeout 600s (Critical UX)
+
+`config.pdf.ocr.request_timeout_secs = 600` (10분/page default). metro-korea.pdf dogfood 증거:
+- page 8 + page 13 에서 Ollama remote 의 slow response → 600s 완전 timeout.
+- 결과: `ms: 600000, chars: 0, skipped: true` emit → 본문 indexed 안 됨 + 20분 cost waste.
+
+**Production impact**: 사용자가 ingest 완료 signal 못 받음, 일부 page 검색 불가.
+
+### §1.4 Bug #13: schema.models single value (UX)
+
+```json
+{
+  "chunker_version": "md-heading-v1",
+  "parser_version": "md-frontmatter-v2",
+  ...
+}
+```
+
+그러나 corpus 안 multi-active:
+- parsers: `md-frontmatter-v2`, `pdf-text-v1`, `code-rust-v1`, `code-python-v1`, `none-v1`.
+- chunkers: `md-heading-v1`, `pdf-page-v1.1`, `code-rust-ast-v1`, `code-python-ast-v1`, `dockerfile-file-v1`, `k8s-manifest-resource-v1`, `manifest-file-v1`, `code-text-paragraph-v1`.
+
+**Impact**: user 가 `kebab schema` 보고 active version 식별 불완전, version cascade audit 시 누락 risk.
+
+### §1.5 Bug #14: empty query silent (Minor UX)
+
+```bash
+kebab search "" --json
+# exit=0, {"hits":[],"next_cursor":null,"schema_version":"search_response.v1"}
+```
+
+empty query (또는 whitespace-only) 가 silent 0 hit return. user mistake → explicit error 가 정합.
+
+---
+
+## §2 Scope + non-scope
+
+### §2.1 Included: 5 bug fix
+
+| Bug | Category | Severity | Fix type |
+|-----|----------|----------|----------|
+| #9 | wire schema | critical | capability flag hardcoded boolean → actual feature check |
+| #10 | config UX | medium | silent fallback → error.v1 with config_not_found |
+| #11 | OCR config | critical | default 600s → 60s timeout |
+| #13 | wire schema | medium | single field → additive array fields (backward compat) |
+| #14 | input validation | minor | empty query silent → error.v1 with invalid_input |
+
+### §2.2 Out of scope
+
+- **Bug #12 (falsified)**: `inspect doc` blocks[].text 가 code parser 에서 "?" placeholder. 근본: `.text` 아님, `.code` field 정상 emit. user workflow 는 `.code` 로 접근 가능 → spec 범위 외.
+- dogfood report §12 의 다른 axis (ranking bias, multi-root caveat) → 별도 phase.
+
+---
+
+## §3 Decisions
+
+### §3.1 Bug #9: capabilities 정정
+
+**Decision**: `schema.rs::capabilities_snapshot()` 의 두 field 를 true 로 update.
+
+```rust
+fn capabilities_snapshot() -> Capabilities {
+    Capabilities {
+        json_mode: true,
+        ingest_progress: true,
+        ingest_cancellation: true,
+        rag_multi_turn: true,
+        search_cache: true,
+        incremental_ingest: true,
+        streaming_ask: true,        // ← WAS FALSE, actual TRUE
+        http_daemon: false,         // ← preserved (not-impl, separate sub-item)
+        mcp_server: true,
+        single_file_ingest: true,   // ← WAS FALSE, actual TRUE
+        bulk_search: true,
+    }
+}
+```
+
+**Rationale**: actual implementation 이 production-grade streaming ask + single-file ingest 지원. schema report 가 reality 와 정합되어야 agent routing 정확함.
+
+### §3.2 Bug #10: config_not_found error
+
+**Decision**: `kebab-config` 가 자체 error type `ConfigNotFound` 정의, `kebab-app::error_wire` 가 classify arm 추가.
+
+Pseudo-code:
+```rust
+// crates/kebab-config/src/lib.rs (또는 적절한 error module)
+#[derive(Debug, thiserror::Error)]
+#[error("config file does not exist: {path}")]
+pub struct ConfigNotFound {
+    pub path: PathBuf,
+}
+
+// Config::load 안:
+pub fn load(opt_path: Option<&Path>) -> anyhow::Result<Config> {
+    match opt_path {
+        Some(p) if !p.exists() => Err(anyhow::Error::new(ConfigNotFound { path: p.to_path_buf() })),
+        Some(p) => Self::from_file(p),
+        None => Self::from_xdg_default_or_defaults(),
+    }
+}
+```
+
+Classify arm in `kebab-app/src/error_wire.rs`:
+```rust
+if let Some(e) = err.downcast_ref::<kebab_config::ConfigNotFound>() {
+    return ErrorV1 {
+        schema_version: ERROR_V1_ID.to_string(),
+        code: "config_not_found".to_string(),
+        message: format!("config file does not exist: {}", e.path.display()),
+        details: json!({ "path": e.path }),
+        hint: Some("verify --config argument; use --config to point to a writable toml file, or omit to use XDG default".to_string()),
+    };
+}
+```
+
+**Exit code**: 2 (config error, not 0 silent).
+
+### §3.3 Bug #11: OCR timeout 60s
+
+**Decision**: `default_pdf_ocr_request_timeout_secs()` → 600 에서 60 으로 감소.
+
+```rust
+fn default_pdf_ocr_request_timeout_secs() -> u64 {
+    60  // 1 min, production-friendly per dogfood evidence
+}
+```
+
+**Doc-comment 추가**:
+```rust
+/// Default OCR request timeout in seconds. Most pages complete in 6-32s.
+/// Set to upper-bound valid throughput; exceeding 60s may indicate
+/// Ollama unavailability or very dense/high-res pages.
+/// Override via [pdf.ocr] request_timeout_secs = N in config.toml.
+```
+
+### §3.4 Bug #13: active_parsers + active_chunkers (additive)
+
+**Decision**: wire schema additive minor — `Models` struct 에 두 배열 추가, 기존 single field 보존 (backward compat). `kebab-store-sqlite` 가 fetch methods 제공.
+
+**Store API** (crates/kebab-store-sqlite/src/lib.rs):
+```rust
+impl SqliteStore {
+    /// SELECT DISTINCT parser_version FROM documents WHERE parser_version IS NOT NULL ORDER BY parser_version
+    pub fn fetch_distinct_parser_versions(&self) -> anyhow::Result<Vec<String>> {
+        let conn = self.conn()?;
+        let mut stmt = conn.prepare(
+            "SELECT DISTINCT parser_version FROM documents
+              WHERE parser_version IS NOT NULL
+              ORDER BY parser_version"
+        )?;
+        let rows = stmt.query_map([], |row| row.get::<_, String>(0))?;
+        let mut out = Vec::new();
+        for r in rows { out.push(r?); }
+        Ok(out)
+    }
+    
+    pub fn fetch_distinct_chunker_versions(&self) -> anyhow::Result<Vec<String>> {
+        let conn = self.conn()?;
+        let mut stmt = conn.prepare(
+            "SELECT DISTINCT chunker_version FROM chunks
+              WHERE chunker_version IS NOT NULL
+              ORDER BY chunker_version"
+        )?;
+        let rows = stmt.query_map([], |row| row.get::<_, String>(0))?;
+        let mut out = Vec::new();
+        for r in rows { out.push(r?); }
+        Ok(out)
+    }
+}
+```
+
+**Models struct** (crates/kebab-app/src/schema.rs):
+```rust
+pub struct Models {
+    /// Deprecated since v0.20.1. Use active_parsers for multi-parser corpus.
+    /// Reports default parser version (markdown path).
+    pub parser_version: String,
+    
+    /// Deprecated since v0.20.1. Use active_chunkers for multi-chunker corpus.
+    pub chunker_version: String,
+    
+    /// All parser versions active in corpus (v0.20.1+). May be empty if corpus is empty.
+    pub active_parsers: Vec<String>,
+    
+    /// All chunker versions active in corpus (v0.20.1+). May be empty if corpus is empty.
+    pub active_chunkers: Vec<String>,
+    
+    pub embedding_version: String,
+    pub prompt_template_version: String,
+    pub index_version: String,
+    pub corpus_revision: u64,
+}
+```
+
+**Computation** (crates/kebab-app/src/schema.rs::collect_models):
+```rust
+let store = open_store_for_stats(cfg)?;
+let active_parsers = store.fetch_distinct_parser_versions().unwrap_or_default();
+let active_chunkers = store.fetch_distinct_chunker_versions().unwrap_or_default();
+
+Ok(Models {
+    parser_version: active_parsers.first().cloned().unwrap_or_else(|| kebab_parse_md::PARSER_VERSION.to_string()),
+    chunker_version: active_chunkers.first().cloned().unwrap_or_else(|| kebab_chunk::md_heading_v1::VERSION_LABEL.to_string()),
+    active_parsers,
+    active_chunkers,
+    ...
+})
+```
+
+**Fallback**: markdown-fallback 유지. 기존 `parser_version` + `chunker_version` hardcode 보존 (backward compat).
+
+### §3.5 Bug #14: empty query validation
+
+**Decision**: `search` 및 `ask` command 모두에 query empty check + error.v1 emit.
+
+**Search command** (crates/kebab-cli/src/main.rs::search arm):
+```rust
+if let Some(q) = query.as_ref() {
+    if q.trim().is_empty() {
+        return Err(anyhow::Error::new(kebab_app::StructuredError(ErrorV1 {
+            schema_version: ERROR_V1_ID.to_string(),
+            code: "invalid_input".to_string(),
+            message: "query is empty; provide a non-empty search term or use --bulk".into(),
+            details: Value::Null,
+            hint: Some("e.g. `kebab search 'rust async'` or `kebab search --bulk < queries.ndjson`".into()),
+        })));
+    }
+}
+```
+
+**Ask command** (crates/kebab-cli/src/main.rs::ask arm):
+```rust
+if query.trim().is_empty() {
+    return Err(anyhow::Error::new(kebab_app::StructuredError(ErrorV1 {
+        schema_version: ERROR_V1_ID.to_string(),
+        code: "invalid_input".to_string(),
+        message: "query is empty; provide a non-empty prompt".into(),
+        details: Value::Null,
+        hint: Some("e.g. `kebab ask 'explain this code'`".into()),
+    })));
+}
+```
+
+Both commands now validate; no silent fallback.
+
+---
+
+## §4 Implementation specification
+
+### §4.1 Files to modify
+
+1. **Bug #9 capability fix**: `crates/kebab-app/src/schema.rs`
+   - line 137–151: `capabilities_snapshot()` — flip `streaming_ask: false` → `true`, `single_file_ingest: false` → `true`.
+   - add test: `capabilities_streaming_ask_matches_cli_surface()`.
+   - add test: `capabilities_single_file_ingest_matches_cli_surface()`.
+
+2. **Bug #10 config_not_found**: Two files
+   - `crates/kebab-config/src/lib.rs`:
+     - Define `ConfigNotFound` error struct (with `#[derive(Debug, thiserror::Error)]`).
+     - Modify `Config::load(opt_path: Option<&Path>)` — path existence check, `return Err(anyhow::Error::new(ConfigNotFound { ... }))`.
+     - add test: `config_load_explicit_nonexistent_path_returns_error()`.
+   - `crates/kebab-app/src/error_wire.rs`:
+     - Add classify arm after existing `ConfigInvalid` case.
+     - Map `kebab_config::ConfigNotFound` → `ErrorV1 { code: "config_not_found", ... }`.
+
+3. **Bug #13 schema.models**: Three components
+   - `crates/kebab-store-sqlite/src/lib.rs`:
+     - Implement `fetch_distinct_parser_versions()` — SQL SELECT DISTINCT on documents.parser_version + ORDER BY.
+     - Implement `fetch_distinct_chunker_versions()` — SQL SELECT DISTINCT on chunks.chunker_version + ORDER BY.
+   - `crates/kebab-app/src/schema.rs`:
+     - Modify `Models` struct — add `active_parsers: Vec<String>`, `active_chunkers: Vec<String>` fields.
+     - Modify computation logic (`collect_models` or equiv) — call store methods, populate arrays, fallback to markdown defaults for single fields.
+     - add test: `schema_models_active_arrays_empty_on_empty_corpus()`.
+     - add test: `schema_models_active_arrays_populated_after_mixed_ingest()`.
+   - `docs/wire-schema/v1/schema.schema.json`:
+     - `Models` object — add `"active_parsers": { "type": "array", "items": { "type": "string" } }`.
+     - add `"active_chunkers": { "type": "array", "items": { "type": "string" } }`.
+     - Mark deprecated in comment: `parser_version` + `chunker_version` (additive, backward compat).
+
+4. **Bug #14 empty query validation**: `crates/kebab-cli/src/main.rs`
+   - search command arm: add `if query.trim().is_empty()` check → error.v1 code=invalid_input.
+   - ask command arm: add identical `if query.trim().is_empty()` check → error.v1 code=invalid_input.
+
+5. **Wire schema v1 doc update**: `docs/wire-schema/v1/`
+   - Update schema doc to note `active_parsers` / `active_chunkers` optional (additive).
+
+6. **Integration**: `integrations/claude-code/kebab/SKILL.md`
+   - Update `schema.models` surface docs — reference new `active_*` arrays for multi-version corpora.
+
+7. **Tests** (new or extended):
+   - `crates/kebab-cli/tests/`: invalid --config path (absolute + relative) → error.v1 + exit≠0.
+   - `crates/kebab-cli/tests/`: empty query (search + ask) → error.v1 code=invalid_input + exit≠0.
+   - `crates/kebab-config/tests/`: config file not found → ConfigNotFound error.
+   - `crates/kebab-app/tests/`: mixed corpus schema — active_parsers/chunkers include all ingested versions.
+
+### §4.2 Regression checks
+
+- Existing 1350 workspace tests: `cargo test --workspace --no-fail-fast -j 1` must pass green.
+- All non-bug capabilities (json_mode, ingest_progress, ingest_cancellation, rag_multi_turn, search_cache, incremental_ingest, mcp_server, bulk_search) stay true.
+- Default config path resolution (no --config) unchanged — silent fallback to XDG only if `--config` not passed.
+- Relative path behavior (cwd-relative, Rust std path::Path::exists()) preserved.
+- Empty corpus → empty `active_parsers` / `active_chunkers` array (not null, not error).
+- Existing hardcoded `parser_version` + `chunker_version` fields continue to report markdown defaults (backward compat).
+- Schema version bump not required (wire schema additive minor, backward compat).
+
+---
+
+## §5 Acceptance criteria
+
+| # | Criterion | Evidence |
+|----|-----------|----------|
+| AC-1 | `kebab schema --json` emit `streaming_ask: true` + `single_file_ingest: true` | `cargo test -p kebab-app capabilities_* -j 4` green |
+| AC-2 | `kebab search "x" --config /nonexistent.toml --json` emit exit≠0 + error.v1 code=config_not_found | `cargo test -p kebab-config config_load_explicit_nonexistent_path_returns_error -j 4` green |
+| AC-3 | `cargo test -p kebab-config pdf_ocr_request_timeout_default_is_60s -j 4` → green | unit test confirms default = 60s (no manual timing) |
+| AC-4 | After mixed ingest (MD + PDF + code), `kebab schema --json` emits both `active_parsers` + `active_chunkers` arrays containing all versions | integration test pass |
+| AC-5 | `kebab search "" --json` and `kebab search "  " --json` both emit exit≠0 + error.v1 code=invalid_input | integration test pass |
+| AC-6 | `kebab ask "" --json` emit exit≠0 + error.v1 code=invalid_input (ask symmetry) | integration test pass |
+| AC-7 | `kebab search "rust" --config nonexistent-relative.toml --json` (relative path) emit exit≠0 + error.v1 code=config_not_found | integration test pass |
+| AC-8 | All 1350+ workspace tests pass; no new failures | `cargo test --workspace --no-fail-fast -j 1` exit=0 |
+| AC-9 | Wire schema backward compat: old clients reading `parser_version` + `chunker_version` still work; `active_*` arrays optional per schema | JSON schema `additionalProperties: false` review |
+| AC-10 | `kebab ask --stream` still works; streaming events emitted (no regression) | manual `kebab ask --stream "explain this" 2>&1 | head -3` |
+
+---
+
+## §6 Risks + resolutions
+
+### Risks
+
+- **R-1** (Bug #10): Relative path `./config.toml` must resolve from cwd, not from binary location. **Resolution**: Rust `std::path::Path::exists()` is cwd-relative; no workaround needed.
+- **R-2** (Bug #13): Empty corpus → empty `active_parsers` / `active_chunkers` array. **Resolution**: Unit test `schema_models_active_arrays_empty_on_empty_corpus()` mandated (AC-4).
+- **R-3** (resolved): `collect_models` uses no cache (every-call re-computation). `active_parsers/chunkers` reflect corpus state at invocation time. If future caching is added, `corpus_revision` increment signals invalidation — document at that time.
+- **R-4** (Bug #14): `ask` command validation — covered by same fix (§3.5 mandates both search + ask).
+- **R-5** (Bug #11): 60s may still timeout on very dense/high-res pages. **Mitigation**: User can override via `config.toml [pdf.ocr] request_timeout_secs = N`. Release notes explicitly call this out.
+
+---
+
+---
+
+## §7 Parent spec deviation (HOTFIXES handoff)
+
+**F-11 MEDIUM finding**: parent spec `2026-04-27-kebab-final-form-design.md` (frozen) specifies PDF OCR request_timeout_secs = 600s (§1000 + §1628 OQ-1, rationale: "CPU 환경 105s 의 5x 여유"). Bug #11 (dogfood evidence) contradicts — 600s causes timeouts; 60s production-optimal.
+
+**Deviation handling**:
+1. Parent spec stays frozen (no edits).
+2. **HOTFIXES entry (executor Step N)**: `tasks/HOTFIXES.md` receives dated entry:
+   ```markdown
+   2026-05-27 — PDF OCR request_timeout_secs default 600s → 60s (v0.20.0 bugfix3 dogfood evidence). Bug #11.
+   ```
+3. **Parent spec cross-link (executor Step N)**: parent spec `2026-04-27-kebab-final-form-design.md` receives inline comment at §1000 (default value code block) or §1628 (OQ-1 paragraph):
+   ```markdown
+   <!-- HOTFIX 2026-05-27: default 60s (Bug #11). See tasks/HOTFIXES.md 2026-05-27 entry. -->
+   ```
+
+**Parent spec invariant**: No changes to parent spec text; only cross-link comment + HOTFIXES.md entry. Frozen design contract preserved.
+
+---
+
+## §8 References
+
+- [Dogfood report](../../../.omc/reviews/2026-05-27-v0.20-final-dogfood-report.md) — 5 bugs discovered + decisions.
+- [Parent spec (frozen contract)](2026-04-27-kebab-final-form-design.md) — §1, §2, §4 (capabilities, error handling, JSON schema, config XDG).
+- `crates/kebab-app/src/schema.rs:137–151` (capabilities_snapshot).
+- `crates/kebab-config/src/lib.rs` (Config::load, default_pdf_ocr_request_timeout_secs).
+- `crates/kebab-app/src/error_wire.rs` (classify ConfigNotFound).
+- `crates/kebab-store-sqlite/src/lib.rs` (fetch_distinct_parser_versions, fetch_distinct_chunker_versions).
+- `crates/kebab-cli/src/main.rs` (search + ask query validation).
+- `docs/wire-schema/v1/schema.schema.json` (Models + Capabilities objects).
+- `tasks/HOTFIXES.md` (2026-05-27 entry, Bug #11 deviation record).