2026-06-23 15:52:35 +00:00
101 changed files with 1201 additions and 111 deletions
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -4751,7 +4751,7 @@ dependencies = [

 [[package]]
 name = "kebab-app"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "base64 0.22.1",
@@ -4799,7 +4799,7 @@ dependencies = [

 [[package]]
 name = "kebab-chunk"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -4817,7 +4817,7 @@ dependencies = [

 [[package]]
 name = "kebab-cli"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "clap",
@@ -4838,7 +4838,7 @@ dependencies = [

 [[package]]
 name = "kebab-config"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "dirs 5.0.1",
@@ -4854,7 +4854,7 @@ dependencies = [

 [[package]]
 name = "kebab-core"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -4868,7 +4868,7 @@ dependencies = [

 [[package]]
 name = "kebab-embed"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -4882,7 +4882,7 @@ dependencies = [

 [[package]]
 name = "kebab-embed-candle"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "candle-core",
@@ -4902,7 +4902,7 @@ dependencies = [

 [[package]]
 name = "kebab-embed-local"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "fastembed",
@@ -4915,7 +4915,7 @@ dependencies = [

 [[package]]
 name = "kebab-embed-ollama"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-config",
@@ -4930,7 +4930,7 @@ dependencies = [

 [[package]]
 name = "kebab-eval"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-app",
@@ -4949,7 +4949,7 @@ dependencies = [

 [[package]]
 name = "kebab-llm"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-core",
@@ -4958,7 +4958,7 @@ dependencies = [

 [[package]]
 name = "kebab-llm-local"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-config",
@@ -4975,7 +4975,7 @@ dependencies = [

 [[package]]
 name = "kebab-mcp"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-app",
@@ -4993,7 +4993,7 @@ dependencies = [

 [[package]]
 name = "kebab-nli"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "hf-hub",
@@ -5008,7 +5008,7 @@ dependencies = [

 [[package]]
 name = "kebab-parse-code"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "gix",
@@ -5031,7 +5031,7 @@ dependencies = [

 [[package]]
 name = "kebab-parse-image"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "ab_glyph",
 "anyhow",
@@ -5059,7 +5059,7 @@ dependencies = [

 [[package]]
 name = "kebab-parse-md"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "kebab-core",
@@ -5076,7 +5076,7 @@ dependencies = [

 [[package]]
 name = "kebab-parse-pdf"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -5091,7 +5091,7 @@ dependencies = [

 [[package]]
 name = "kebab-rag"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -5113,7 +5113,7 @@ dependencies = [

 [[package]]
 name = "kebab-search"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "globset",
@@ -5132,7 +5132,7 @@ dependencies = [

 [[package]]
 name = "kebab-source-fs"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -5150,7 +5150,7 @@ dependencies = [

 [[package]]
 name = "kebab-store-sqlite"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "blake3",
@@ -5170,7 +5170,7 @@ dependencies = [

 [[package]]
 name = "kebab-store-vector"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "arrow",
@@ -5194,7 +5194,7 @@ dependencies = [

 [[package]]
 name = "kebab-tui"
-version = "0.28.0"
+version = "0.29.0"
 dependencies = [
 "anyhow",
 "crossterm",
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -32,7 +32,7 @@ edition       = "2024"
 rust-version  = "1.85"
 license       = "MIT OR Apache-2.0"
 repository    = "https://github.com/altair823/kebab"
-version       = "0.28.0"   # v0.28.0 — config 스키마 v2→v3 재편: 미디어 형식 설정을 `[ingest.*]` 우산으로 통합(`[indexing]`→`[ingest]` 스칼라, `[chunking]`/`[image.ocr]`/`[image.caption]`/`[pdf.ocr]`→`[ingest.*]`). 기존 v2 파일은 load 시 메모리 자동 변환(디스크 미변경), 파일 갱신은 `kebab config migrate`(값·주석 보존). env 이름(LHS) 100% 보존 + RHS 만 새 경로, 신규 `KEBAB_PDF_OCR_{DET_MODEL,REC_MODEL,DICT,SCORE_THRESH,UNCLIP_RATIO,MAX_BOXES}`. `ingest_config_signature` 바이트 불변(재색인 0). PdfOcrCfg paddle 대칭 키. 신규 인터페이스(config 레이아웃 rename + env 추가) → minor. — CLAUDE.md §Release
+version       = "0.29.0"   # v0.29.0 — provenance 출처 필터: `[[workspace.sources]]` 멀티소스 + 검색 `--source <id>` / `--source-type <type>`(lexical+vector 두 site, OR). `documents.source_id` 컬럼(V014, additive·DEFAULT 'default'·재색인 0) + config v3→v4 migration(`step_3_to_4`, 단일 root→implicit `default` source 미러, 멱등). per-source `trust_level`/`source_type` 기본값(우선순위 frontmatter > source 기본값 > Primary). 단일 root 사용자 무영향. 설계 근거: 전역 trust 곱셈가중(weighted-RRF)은 A/B 반증(incident MRR 절벽), 출처 필터가 see-saw 없는 레버. 신규 CLI flag + config 키 + migration → minor. — CLAUDE.md §Release

 # pre-v0.18 workspace-wide cleanup: enable clippy::pedantic group with
 # intentional allow-list. The allowed lints are either cosmetic (doc style),
--- a/HANDOFF.md
+++ b/HANDOFF.md
@@ -35,6 +35,7 @@ P0~P5 직렬. P6~P9 P5 이후 병렬 가능.

 머지 후 발견된 모든 deviation / hotfix 의 dated 로그는 [tasks/HOTFIXES.md](tasks/HOTFIXES.md). 본 요약은 \"누군가가 인수받을 때 알아두면 시간을 많이 절약하는\" 항목만:

+- **2026-06-21 provenance 출처 필터: `[[workspace.sources]]` 멀티소스 + `--source`/`--source-type`** — v0.29.0. 혼합 출처 KB(위키+jira 등)에서 색인은 전부 하되 질의 시 출처로 좁히는 레버. config `[[workspace.sources]]`(각 id/root/trust_level/source_type) + `documents.source_id` 컬럼(V014, additive, 재색인 0) + config v3→v4 migration(`step_3_to_4`, 단일 root→implicit `default` source, 멱등) + 검색 `--source <id>` / `--source-type <type>`(lexical+vector 두 site, OR). trust precedence = frontmatter > per-source 기본값 > Primary. **설계 근거**: 전역 trust 곱셈가중(weighted-RRF)은 A/B 에서 반증(θ=0.85 만으로 incident MRR 0.918→0.340 절벽) — 필터가 see-saw 없는 올바른 레버. 도그푸딩(620 doc, jira400+wiki220): `--source wiki` concept 0.780→0.810, `--source jira` incident 0.918→0.975. **follow-up**: MCP search 필터 미노출 · `kebab list` source_id 미표시 · RAG provenance 라벨 미구현. 자세한 내용: `tasks/HOTFIXES.md` (2026-06-21).
 - **2026-06-04 PP-OCRv5 ONNX Rust 네이티브 OCR** — v0.27.0. `[image.ocr] engine = "paddle-onnx"` 로 PP-OCRv5(검출+인식) ONNX 를 in-process(`ort` =2.0.0-rc.9) 실행 — Python 런타임/원격 호출 없이 큰 페이지 CPU <4초(Ollama vision ~50초 대비). default 는 여전히 `"ollama-vision"`. 후처리(min-area rect/unclip)는 pure-Rust. **함정**: unclip 은 corner 를 centroid 에서 방사 확장하면 안 되고 edge 별 polygon offset 이어야 함(방사 확장 시 wide/short 텍스트 박스 높이가 안 커져 글자 윗부분 잘림 → ㄷ→ㄴ, e2e CER 0.26). 수정 후 CER 0.005. 모델 ONNX 는 `crates/kebab-parse-image/assets/paddleocr-onnx/`(LFS). 자세한 내용: `tasks/HOTFIXES.md` (2026-06-04 PP-OCRv5 ONNX), spec/plan `docs/superpowers/{specs,plans}/2026-06-04-rust-native-ocr-*.md`.
 - **2026-06-03 ingest 설정 변경 자동 재색인** — v0.26.2. ingest 산출에 영향 주는 설정(청킹/이미지 OCR·caption/pdf.ocr/`[ingest.code]`)을 변경하면 `--force-reingest` 없이 영향 자산만 자동 재색인. 그 설정들의 결정적 서명(`ingest_config_signature`)을 effective parser_version(skip 비교 + 저장 doc 필드 양쪽)에 폴딩 → 다음 ingest 비교가 mismatch. 비산출 설정(search/rag/ui/log + max_pixels/languages/timeout)은 제외(과도 무효화 회피), doc_id 는 base 로 안정 유지. **업그레이드 후 첫 ingest 는 전 자산 1회 재색인**(저장된 상수 parser_version ≠ 새 composite; embedding 은 V012 캐시 히트). 결과 포맷·CLI·wire 불변(내부 skip 판정 정정). 자세한 내용: `tasks/HOTFIXES.md` (2026-06-03 ingest 설정 변경 자동 재색인), spec/plan `docs/superpowers/{specs,plans}/2026-06-03-*invalidation*.md`.
 - **2026-06-03 ingest 진행 로그 개선** — v0.26.1. 이미지/PDF + OCR/caption on 볼트 ingest 가 "멈춘 듯" 보이던 문제 해소: TTY 진행바에 현재 파일명 + 느린 phase(ocr/caption/embed)+모델명 + 경과초 `(Ns)` heartbeat, 종료 시 최장 소요 파일 top-5 요약. 신규 wire `asset_phase{idx,total,phase,model}` + `asset_timings.ocr_ms`/`caption_ms`(additive, `ingest_progress.v1` 유지, serde default 0). 이미지·PDF 경로도 `asset_timings` emit(이전 markdown 만). 기본 동작 불변. 자세한 내용: `tasks/HOTFIXES.md` (2026-06-03 ingest 진행 로그), spec/plan `docs/superpowers/{specs,plans}/2026-06-03-ingest-log-improve-*.md`.
--- a/README.md
+++ b/README.md
@@ -86,7 +86,7 @@ Markdown · PDF · 이미지(OCR + caption) · 소스코드(Rust/Python/TS/JS/Go
 | `kebab ingest [<path>]` | 워크스페이스 스캔 후 새/변경 문서 색인 (idempotent · incremental, `--force-reingest` 로 강제 재처리). 미지원 확장자는 자동 skip. 진행바는 현재 **파일명** · 느린 **phase(ocr/caption/embed)+모델명** · **경과초**`(Ns)` · 문서별 청크 수 · phase별 소요시간(parse/chunk/ocr/caption/embed/store)을 표시하고, 종료 시 **최장 소요 파일 top-5** 를 요약한다 (`--json` 은 `asset_phase`/`asset_chunked`/`asset_timings` 이벤트로, 사람용 요약은 미출력) |
 | `kebab ingest-file <path>` | 단일 파일 ingest (workspace 외부 가능 — `_external/` 로 deterministic copy) |
 | `kebab ingest-stdin --title <T>` | stdin 의 markdown 본문 ingest |
-| `kebab search --mode {lexical,vector,hybrid} "<query>" [flags]` | 검색 (default hybrid = RRF fusion, citation 포함). 필터/budget flag 는 `--help` |
+| `kebab search --mode {lexical,vector,hybrid} "<query>" [flags]` | 검색 (default hybrid = RRF fusion, citation 포함). 출처 필터 `--source <id>` (`[[workspace.sources]]` id) · `--source-type {markdown,note,paper,reference,inbox}` (둘 다 repeatable/comma-sep, OR). 그 외 필터/budget flag 는 `--help` |
 | `kebab ask "<query>" [flags]` | RAG 답변 + 근거 인용 (Ollama 필요). `--session` (multi-turn) · `--stream` · `--multi-hop` |
 | `kebab list docs` | 색인된 문서 목록 |
 | `kebab inspect doc <id>` / `inspect chunk <id>` | raw record 보기 |
@@ -108,6 +108,19 @@ Markdown · PDF · 이미지(OCR + caption) · 소스코드(Rust/Python/TS/JS/Go
 [workspace]
 root = "~/KnowledgeBase"   # 색인할 폴더. 절대 / tilde / env / 상대 경로 가능.
                          # 상대 경로의 base 는 config.toml 위치 (cwd 무관).
+                          # 단일 root 는 implicit `default` source 로 정규화된다.
+
+# 멀티소스 (선택) — 출처별로 검색을 좁히려면 root 대신 명명 source 를 선언한다.
+# 각 source 의 id 가 모든 문서에 stamp 되고, `kebab search --source <id>` 로 필터.
+# trust_level / source_type 은 frontmatter 가 없을 때의 source 기본값
+# (우선순위: frontmatter > source 기본값 > 하드코딩 Primary/Markdown).
+# [[workspace.sources]]
+# id = "notes"
+# root = "~/KnowledgeBase/notes"
+# [[workspace.sources]]
+# id = "jira"
+# root = "~/exports/jira"
+# trust_level = "secondary"   # 낮은 신뢰 출처 — `--trust-min primary` 로 배제 가능.

 [models.embedding]
 provider = "fastembed"            # "fastembed"(기본, onnxruntime) / "candle"(순수 Rust)
--- a/crates/kebab-app/src/app.rs
+++ b/crates/kebab-app/src/app.rs
@@ -727,8 +727,7 @@ impl App {
        // Load (or create) the session header.
        let now_unix = SystemTime::now()
            .duration_since(UNIX_EPOCH)
-            .map(|d| d.as_secs() as i64)
-            .unwrap_or(0);
+            .map_or(0, |d| d.as_secs() as i64);
        let existing = self.sqlite.get_session(session_id)?;
        let prior_turns = match &existing {
            Some(_) => self.sqlite.list_turns(session_id)?,
@@ -1111,7 +1110,7 @@ fn trim_to_chars(s: &str, n: usize) -> String {
 /// terminates early) rather than panic in the budget loop.
 fn estimate_chars(hits: &[SearchHit]) -> usize {
    hits.iter()
-        .map(|h| serde_json::to_string(h).map(|s| s.len()).unwrap_or(0))
+        .map(|h| serde_json::to_string(h).map_or(0, |s| s.len()))
        .sum()
 }

--- a/crates/kebab-app/src/bulk.rs
+++ b/crates/kebab-app/src/bulk.rs
@@ -206,6 +206,8 @@ fn parse_one(raw: &Value) -> Result<(SearchQuery, SearchOpts), String> {
        doc_id,
        repo: vec![],
        code_lang: vec![],
+        source_type: vec![],
+        source_id: vec![],
    };

    let opts = SearchOpts {
--- a/crates/kebab-app/src/lib.rs
+++ b/crates/kebab-app/src/lib.rs
@@ -49,7 +49,8 @@ use kebab_core::{
    Answer, Block, CanonicalDocument, Chunk, ChunkId, ChunkPolicy, Chunker, ChunkerVersion,
    DocFilter, DocSummary, DocumentId, DocumentStore, Embedder, EmbeddingInput, EmbeddingKind,
    ExtractContext, IngestReport, Lang, LanguageModel, MediaType, ParserVersion, RawAsset,
-    SearchHit, SearchQuery, SourceScope, SourceUri, VectorRecord, VectorStore,
+    SearchHit, SearchQuery, SourceScope, SourceType, SourceUri, TrustLevel, VectorRecord,
+    VectorStore,
 };
 use kebab_llm_local::OllamaLanguageModel;
 use kebab_parse_image::{
@@ -304,7 +305,12 @@ pub fn ingest_with_config_opts(
            0
        });

-    // Walk the workspace.
+    // Walk the workspace. `[[workspace.sources]]`: when the caller did not
+    // pin an explicit `scope.root` (the normal `kebab ingest` path), iterate
+    // over every configured source — each scanned with its own root + exclude
+    // and tagged with its `id` + default trust. When `scope.root` IS pinned
+    // (single-file ingest, `--root` override), scan that one root as the
+    // implicit `default` source — preserving pre-multi-source behavior.
    crate::ingest_progress::emit(
        progress,
        crate::ingest_progress::IngestEvent::ScanStarted {
@@ -313,9 +319,50 @@ pub fn ingest_with_config_opts(
    );
    let connector =
        FsSourceConnector::new(&app.config).context("kb-app::ingest: build FsSourceConnector")?;
-    let (assets, fs_skips) = connector
-        .scan_with_skips(&scope)
-        .context("kb-app::ingest: scan workspace")?;
+
+    // Per-source scan plan: (source_id, source_trust, scan_scope).
+    let scan_plan: Vec<(String, Option<TrustLevel>, SourceScope)> =
+        if scope.root.as_os_str().is_empty() && scope.include.is_empty() {
+            app.config
+                .resolved_sources()
+                .into_iter()
+                .map(|s| {
+                    let scan_scope = SourceScope {
+                        root: s.root,
+                        include: scope.include.clone(),
+                        exclude: s.exclude,
+                    };
+                    (s.id, s.trust_level, scan_scope)
+                })
+                .collect()
+        } else {
+            // Explicit-root / single-file / include-restricted ingest: one
+            // ad-hoc `default` source rooted at the pinned scope.
+            vec![(
+                kebab_config::DEFAULT_SOURCE_ID.to_string(),
+                None,
+                scope.clone(),
+            )]
+        };
+
+    // Accumulate assets across sources + a per-path lookup of which source
+    // (id + trust) each asset came from. workspace_path is unique per asset
+    // within a scan; on the rare overlap across sources, last-write-wins
+    // (sources should not share roots — a config smell, not enforced).
+    let mut assets: Vec<RawAsset> = Vec::new();
+    let mut source_by_path: std::collections::HashMap<String, (String, Option<TrustLevel>)> =
+        std::collections::HashMap::new();
+    let mut fs_skips = kebab_source_fs::FsScanSkips::default();
+    for (sid, strust, scan_scope) in &scan_plan {
+        let (src_assets, src_skips) = connector
+            .scan_with_skips(scan_scope)
+            .with_context(|| format!("kb-app::ingest: scan source `{sid}`"))?;
+        for a in &src_assets {
+            source_by_path.insert(a.workspace_path.0.clone(), (sid.clone(), *strust));
+        }
+        assets.extend(src_assets);
+        fs_skips.merge(src_skips);
+    }
    crate::ingest_progress::emit(
        progress,
        crate::ingest_progress::IngestEvent::ScanCompleted {
@@ -468,6 +515,14 @@ pub fn ingest_with_config_opts(
                media: crate::ingest_progress::media_label(&asset.media_type).to_string(),
            },
        );
+        // `[[workspace.sources]]`: resolve which source this asset came from.
+        // Missing only if an asset slipped in outside the scan plan (defensive
+        // — fall back to the implicit `default` source).
+        let (source_id, source_trust) = source_by_path
+            .get(&asset.workspace_path.0)
+            .map_or((kebab_config::DEFAULT_SOURCE_ID, None), |(id, trust)| {
+                (id.as_str(), *trust)
+            });
        let item = ingest_one_asset(
            &app,
            &asset,
@@ -478,6 +533,8 @@ pub fn ingest_with_config_opts(
            embedder.as_ref(),
            vector_store.as_ref(),
            &existing_doc_ids,
+            source_id,
+            source_trust,
            &image_pipeline,
            force_reingest,
            pdf_ocr_engine.as_deref(),
@@ -738,8 +795,8 @@ pub fn ingest_with_config_opts(
        if let Ok(mut w) = lw.lock() {
            let run_id = w.run_id().to_string();
            let ms_samples = ocr_ms_samples.lock().map(|v| v.clone()).unwrap_or_default();
-            let pages = ocr_pages_cnt.lock().map(|v| *v).unwrap_or(0);
-            let failures = ocr_failures_cnt.lock().map(|v| *v).unwrap_or(0);
+            let pages = ocr_pages_cnt.lock().map_or(0, |v| *v);
+            let failures = ocr_failures_cnt.lock().map_or(0, |v| *v);
            let summary = crate::ingest_log::IngestSummary::new(
                crate::ingest_log::now_ts(),
                run_id,
@@ -1173,6 +1230,11 @@ fn ingest_one_asset(
    embedder: Option<&Arc<dyn Embedder + Send + Sync>>,
    vector_store: Option<&Arc<kebab_store_vector::LanceVectorStore>>,
    existing_doc_ids: &std::collections::HashSet<String>,
+    // `[[workspace.sources]]`: id of the source this asset belongs to (stamped
+    // onto `documents.source_id`) + that source's default trust level
+    // (markdown frontmatter overrides it).
+    source_id: &str,
+    source_trust: Option<TrustLevel>,
    image_pipeline: &ImagePipeline<'_>,
    force_reingest: bool,
    pdf_ocr_engine: Option<&dyn OcrEngine>,
@@ -1206,6 +1268,7 @@ fn ingest_one_asset(
                embedder,
                vector_store,
                existing_doc_ids,
+                source_id,
                image_pipeline,
                force_reingest,
                progress,
@@ -1221,6 +1284,7 @@ fn ingest_one_asset(
                embedder,
                vector_store,
                existing_doc_ids,
+                source_id,
                force_reingest,
                pdf_ocr_engine,
                progress,
@@ -1263,6 +1327,7 @@ fn ingest_one_asset(
                existing_doc_ids,
                force_reingest,
                lang.as_str(),
+                source_id,
            );
        }
        // p10-1A-2: non-Rust Code, Audio, and Other are not yet wired;
@@ -1338,7 +1403,7 @@ fn ingest_one_asset(
    let bytes = std::fs::read(&path)
        .with_context(|| format!("read asset bytes from {}", path.display()))?;

-    let body_hints = build_body_hints(asset);
+    let body_hints = build_body_hints(asset, Some(source_id), source_trust);

    // Frontmatter — `parse_frontmatter` returns Ok even on malformed
    // frontmatter (warnings are surfaced through the `Vec<Warning>`).
@@ -1572,6 +1637,7 @@ fn ingest_one_image_asset(
    embedder: Option<&Arc<dyn Embedder + Send + Sync>>,
    vector_store: Option<&Arc<kebab_store_vector::LanceVectorStore>>,
    existing_doc_ids: &std::collections::HashSet<String>,
+    source_id: &str,
    image_pipeline: &ImagePipeline<'_>,
    force_reingest: bool,
    progress: Option<&std::sync::mpsc::Sender<crate::ingest_progress::IngestEvent>>,
@@ -1646,6 +1712,9 @@ fn ingest_one_image_asset(
    // `image-meta-v1`, which already fixed doc_id). Skip compare + stored
    // field must agree for next-run detection.
    canonical.parser_version = eff_parser_version.clone();
+    // `[[workspace.sources]]`: stamp the owning source id (image extractor
+    // leaves it None).
+    canonical.metadata.source_id = Some(source_id.to_string());
    let parse_ms = u64::try_from(t_parse.elapsed().as_millis()).unwrap_or(u64::MAX);

    // 2 + 3. Apply OCR / caption when their adapters exist. Both are
@@ -2157,6 +2226,7 @@ fn ingest_one_pdf_asset(
    embedder: Option<&Arc<dyn Embedder + Send + Sync>>,
    vector_store: Option<&Arc<kebab_store_vector::LanceVectorStore>>,
    existing_doc_ids: &std::collections::HashSet<String>,
+    source_id: &str,
    force_reingest: bool,
    pdf_ocr_engine: Option<&dyn OcrEngine>,
    progress: Option<&std::sync::mpsc::Sender<crate::ingest_progress::IngestEvent>>,
@@ -2224,6 +2294,9 @@ fn ingest_one_pdf_asset(
    // v0.26.2: store the composite parser_version (base `pdf-text-v1` already
    // fixed doc_id) so the next run's skip compare matches.
    canonical.parser_version = eff_parser_version.clone();
+    // `[[workspace.sources]]`: stamp the owning source id (pdf extractor
+    // leaves it None).
+    canonical.metadata.source_id = Some(source_id.to_string());
    let parse_ms = u64::try_from(t_parse.elapsed().as_millis()).unwrap_or(u64::MAX);

    // v0.20 sub-item 1: post-extract OCR enrichment (PR #187 registry
@@ -2523,6 +2596,7 @@ fn ingest_one_code_asset(
    existing_doc_ids: &std::collections::HashSet<String>,
    force_reingest: bool,
    code_lang: &str, // <-- NEW (p10-1b Task D)
+    source_id: &str,
 ) -> anyhow::Result<kebab_core::IngestItem> {
    let path = match &asset.source_uri {
        SourceUri::File(p) => p.clone(),
@@ -2679,6 +2753,11 @@ fn ingest_one_code_asset(
        }
    };

+    // `[[workspace.sources]]`: stamp the owning source id on the synthesized /
+    // extracted code doc (covers both Tier 1 extract_for and Tier 2/3
+    // synthesize paths — neither knows the source id).
+    canonical.metadata.source_id = Some(source_id.to_string());
+
    // p10-1b Task D/G/J/L: chunker per-lang.
    // p10-3: track whether the extract stage already fell back to Tier 3.
    // Tier 2 langs already have "none-v1" parser_version normally, so exclude them
@@ -2898,7 +2977,7 @@ fn synthesize_tier2_document(
    use anyhow::Context as _;
    use kebab_core::{
        BlockId, CodeBlock, CommonBlock, Lang, Metadata, Provenance, ProvenanceEvent,
-        ProvenanceKind, SourceSpan, SourceType, TrustLevel, id_for_block, id_for_doc,
+        ProvenanceKind, SourceSpan, id_for_block, id_for_doc,
    };

    let text = std::str::from_utf8(bytes)
@@ -2986,6 +3065,10 @@ fn synthesize_tier2_document(
        git_branch,
        git_commit,
        code_lang: Some(code_lang.to_string()),
+        // `[[workspace.sources]]`: stamped by the caller
+        // (`ingest_one_code_asset`) post-build so Tier 1 (extract_for) and
+        // Tier 2/3 (this synthesizer) share one code path.
+        source_id: None,
    };

    tracing::debug!(
@@ -3044,12 +3127,20 @@ fn count_lines_in(bytes: &[u8]) -> u32 {
 /// overhead for large workspaces and the source-of-truth timestamps
 /// are written into the document's frontmatter when the user wants
 /// authoritative values.
-fn build_body_hints(asset: &RawAsset) -> BodyHints {
+fn build_body_hints(
+    asset: &RawAsset,
+    source_id: Option<&str>,
+    source_trust: Option<TrustLevel>,
+) -> BodyHints {
    BodyHints {
        first_h1: None,
        fs_ctime: asset.discovered_at,
        fs_mtime: asset.discovered_at,
        fallback_lang: None,
+        // `[[workspace.sources]]`: stamp the owning source id + inject the
+        // per-source default trust level (frontmatter still overrides it).
+        source_id: source_id.map(str::to_string),
+        fallback_trust_level: source_trust,
    }
 }

--- a/crates/kebab-app/src/reset.rs
+++ b/crates/kebab-app/src/reset.rs
@@ -114,7 +114,7 @@ pub fn estimate_size_bytes(paths: &[PathBuf]) -> u64 {
            if ft.is_dir() {
                total += walk(&e.path());
            } else if ft.is_file() {
-                total += e.metadata().map(|m| m.len()).unwrap_or(0);
+                total += e.metadata().map_or(0, |m| m.len());
            }
        }
        total
--- a/crates/kebab-app/tests/common/mod.rs
+++ b/crates/kebab-app/tests/common/mod.rs
@@ -51,7 +51,7 @@ impl TestEnv {
        std::fs::create_dir_all(&model_dir).unwrap();

        let mut config = Config::defaults();
-        config.workspace.root = workspace_root.to_string_lossy().into_owned();
+        config.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
        // Drop the ".obsidian" / "node_modules" excludes — they bring
        // in nothing useful for fixtures and just hide debugging.
        config.workspace.exclude.clear();
--- a/crates/kebab-app/tests/ingest_file.rs
+++ b/crates/kebab-app/tests/ingest_file.rs
@@ -14,7 +14,7 @@ fn ingest_file_copies_external_md_and_reports_new() {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
@@ -53,7 +53,7 @@ fn ingest_file_idempotent_on_second_call() {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
@@ -78,7 +78,7 @@ fn ingest_file_errors_on_missing_path() {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
@@ -97,7 +97,7 @@ fn ingest_file_errors_on_unsupported_extension() {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-app/tests/ingest_log_smoke.rs
+++ b/crates/kebab-app/tests/ingest_log_smoke.rs
@@ -17,7 +17,7 @@ fn minimal_config(workspace: &std::path::Path, log_dir: &std::path::Path) -> Con
    std::fs::create_dir_all(&model_dir).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = model_dir.to_string_lossy().into_owned();
@@ -130,7 +130,7 @@ fn ingest_log_disabled_emits_no_file() {
    std::fs::create_dir_all(&model_dir).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = model_dir.to_string_lossy().into_owned();
--- a/crates/kebab-app/tests/ingest_progress.rs
+++ b/crates/kebab-app/tests/ingest_progress.rs
@@ -192,7 +192,7 @@ fn pdf_ocr_progress_emits_started_finished_events() {
    std::fs::create_dir_all(&data_dir).expect("create data dir");

    let mut config = kebab_config::Config::defaults();
-    config.workspace.root = workspace.to_string_lossy().into_owned();
+    config.workspace.root = Some(workspace.to_string_lossy().into_owned());
    config.storage.data_dir = data_dir.to_string_lossy().into_owned();
    config.models.embedding.provider = "none".to_string();
    config.models.embedding.dimensions = 0;
--- a/crates/kebab-app/tests/ingest_stdin.rs
+++ b/crates/kebab-app/tests/ingest_stdin.rs
@@ -12,7 +12,7 @@ fn fresh_cfg(dir: &std::path::Path) -> Config {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
@@ -34,7 +34,7 @@ fn ingest_stdin_writes_frontmatter_and_reports_new() {
    assert_eq!(report.new, 1, "{report:?}");

    // _external/ contains exactly one .md file with frontmatter.
-    let ext_dir = std::path::PathBuf::from(&cfg.workspace.root).join("_external");
+    let ext_dir = cfg.resolve_workspace_root().join("_external");
    let entries: Vec<_> = fs::read_dir(&ext_dir)
        .unwrap()
        .filter_map(std::result::Result::ok)
@@ -56,7 +56,7 @@ fn ingest_stdin_without_source_uri() {
        kebab_app::ingest_stdin_with_config(cfg.clone(), "## Body", "Title", None).unwrap();
    assert_eq!(report.new, 1);

-    let ext_dir = std::path::PathBuf::from(&cfg.workspace.root).join("_external");
+    let ext_dir = cfg.resolve_workspace_root().join("_external");
    let entries: Vec<_> = fs::read_dir(&ext_dir)
        .unwrap()
        .filter_map(std::result::Result::ok)
--- a/crates/kebab-app/tests/schema_active_versions.rs
+++ b/crates/kebab-app/tests/schema_active_versions.rs
@@ -6,7 +6,7 @@ use kebab_core::SourceScope;

 fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path) -> Config {
    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
--- a/crates/kebab-app/tests/schema_report.rs
+++ b/crates/kebab-app/tests/schema_report.rs
@@ -8,7 +8,7 @@ use kebab_core::SourceScope;

 fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path) -> Config {
    let mut config = Config::defaults();
-    config.workspace.root = workspace_root.to_string_lossy().into_owned();
+    config.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    config.workspace.exclude.clear();
    config.storage.data_dir = data_dir.to_string_lossy().into_owned();
    config.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
--- a/crates/kebab-app/tests/skip_reason.rs
+++ b/crates/kebab-app/tests/skip_reason.rs
@@ -9,7 +9,7 @@ use common::TestEnv;
 #[test]
 fn unsupported_extension_skip_carries_warning_and_is_aggregated() {
    let env = TestEnv::lexical_only();
-    let workspace_root = std::path::PathBuf::from(&env.config.workspace.root);
+    let workspace_root = env.config.resolve_workspace_root();
    std::fs::write(workspace_root.join("legacy.docx"), b"unsupported").unwrap();
    std::fs::write(workspace_root.join("Makefile"), b"unsupported").unwrap();

--- a/crates/kebab-chunk/src/code_c_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_c_ast_v1.rs
@@ -242,6 +242,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("c".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_cpp_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_cpp_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("cpp".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_go_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_go_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("go".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_java_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_java_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("java".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_js_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_js_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("javascript".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_kotlin_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_kotlin_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("kotlin".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_python_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_python_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("python".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_rust_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_rust_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("rust".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/code_ts_ast_v1.rs
+++ b/crates/kebab-chunk/src/code_ts_ast_v1.rs
@@ -244,6 +244,7 @@ mod tests {
                git_branch: Some("main".into()),
                git_commit: Some("0".repeat(40)),
                code_lang: Some("typescript".into()),
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: pv,
--- a/crates/kebab-chunk/src/md_heading_v1.rs
+++ b/crates/kebab-chunk/src/md_heading_v1.rs
@@ -450,6 +450,7 @@ mod tests {
                git_branch: None,
                git_commit: None,
                code_lang: None,
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version: kebab_core::ParserVersion("test-parser-0".into()),
--- a/crates/kebab-chunk/src/pdf_page_v1.rs
+++ b/crates/kebab-chunk/src/pdf_page_v1.rs
@@ -355,6 +355,7 @@ mod tests {
                git_branch: None,
                git_commit: None,
                code_lang: None,
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version,
@@ -533,6 +534,7 @@ mod tests {
                git_branch: None,
                git_commit: None,
                code_lang: None,
+                source_id: None,
            },
            provenance: Provenance { events: vec![] },
            parser_version,
--- a/crates/kebab-chunk/tests/code_c_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_c_ast_snapshot.rs
@@ -111,6 +111,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("c".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_cpp_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_cpp_ast_snapshot.rs
@@ -118,6 +118,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("cpp".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_go_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_go_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("go".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_java_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_java_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("java".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_js_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_js_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("javascript".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_kotlin_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_kotlin_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("kotlin".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_python_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_python_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("python".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_rust_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_rust_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("rust".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_text_paragraph_v1.rs
+++ b/crates/kebab-chunk/tests/code_text_paragraph_v1.rs
@@ -68,6 +68,7 @@ fn text_doc(lang: &str, text: &str) -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some(lang.into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/code_ts_ast_snapshot.rs
+++ b/crates/kebab-chunk/tests/code_ts_ast_snapshot.rs
@@ -136,6 +136,7 @@ fn fixed_doc() -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("typescript".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/dockerfile_file_v1.rs
+++ b/crates/kebab-chunk/tests/dockerfile_file_v1.rs
@@ -67,6 +67,7 @@ fn dockerfile_doc(dockerfile_text: &str) -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("dockerfile".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/k8s_manifest_resource_v1.rs
+++ b/crates/kebab-chunk/tests/k8s_manifest_resource_v1.rs
@@ -67,6 +67,7 @@ fn yaml_doc(yaml_text: &str) -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some("yaml".into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-chunk/tests/long_section_snapshot.rs
+++ b/crates/kebab-chunk/tests/long_section_snapshot.rs
@@ -58,6 +58,8 @@ fn long_section_chunks_snapshot() {
        fs_ctime: asset.discovered_at,
        fs_mtime: asset.discovered_at,
        fallback_lang: Some("en".into()),
+        source_id: None,
+        fallback_trust_level: None,
    };
    let (metadata, fm_span, _fm_warns) =
        parse_frontmatter(&bytes, &hints).expect("frontmatter parses");
@@ -133,6 +135,8 @@ fn long_section_chunks_are_deterministic() {
        fs_ctime: asset.discovered_at,
        fs_mtime: asset.discovered_at,
        fallback_lang: Some("en".into()),
+        source_id: None,
+        fallback_trust_level: None,
    };

    let policy = ChunkPolicy {
--- a/crates/kebab-chunk/tests/manifest_file_v1.rs
+++ b/crates/kebab-chunk/tests/manifest_file_v1.rs
@@ -67,6 +67,7 @@ fn manifest_doc(lang: &str, manifest_text: &str) -> CanonicalDocument {
            git_branch: Some("main".into()),
            git_commit: Some("0".repeat(40)),
            code_lang: Some(lang.into()),
+            source_id: None,
        },
        provenance: Provenance { events: vec![] },
        parser_version: pv,
--- a/crates/kebab-cli/src/main.rs
+++ b/crates/kebab-cli/src/main.rs
@@ -193,6 +193,31 @@ enum Cmd {
        )]
        code_lang: Vec<String>,

+        /// Phase-2: filter by document source_type
+        /// (`markdown`, `note`, `paper`, `reference`, `inbox`).
+        /// Repeatable or comma-separated. Empty = no filter.
+        /// The clean source/provenance lever for mixed-source KBs.
+        #[arg(
+            long = "source-type",
+            value_name = "TYPE",
+            num_args = 1,
+            value_delimiter = ','
+        )]
+        source_type: Vec<String>,
+
+        /// [[workspace.sources]]: filter by source id — the `id` of the
+        /// `[[workspace.sources]]` entry a document was ingested from
+        /// (e.g. `default`, `notes`, `code`). Repeatable or
+        /// comma-separated. Empty = no filter. The named-source
+        /// provenance lever for multi-source KBs.
+        #[arg(
+            long = "source",
+            value_name = "ID",
+            num_args = 1,
+            value_delimiter = ','
+        )]
+        source: Vec<String>,
+
        /// p9-fb-37: emit pre-fusion lexical / vector / RRF candidate
        /// lists + per-stage timing in the response. Bypasses cache
        /// (debug intent — fresh run guaranteed). Requires embeddings
@@ -615,12 +640,18 @@ fn run(cli: &Cli) -> anyhow::Result<()> {
            force_reingest,
        } => {
            let cfg = kebab_config::Config::load(cli.config.as_deref())?;
-            let scope = kebab_core::SourceScope {
-                root: root
-                    .clone()
-                    .unwrap_or_else(|| PathBuf::from(&cfg.workspace.root)),
+            // [[workspace.sources]]: when the user passes `--root <dir>` we pin
+            // that single root (one ad-hoc `default` source). Otherwise we
+            // leave `scope.root` EMPTY so the app iterates every configured
+            // source (`config.resolved_sources()`); a bare empty scope.exclude
+            // is fine because each source carries its own merged exclude.
+            let scope = match root.clone() {
+                Some(r) => kebab_core::SourceScope {
+                    root: r,
                    exclude: cfg.workspace.exclude.clone(),
                    ..Default::default()
+                },
+                None => kebab_core::SourceScope::default(),
            };

            // p9-fb-02: spawn the progress display on a background
@@ -629,8 +660,7 @@ fn run(cli: &Cli) -> anyhow::Result<()> {
            // call returns, the `Sender` drops and the display thread
            // sees `recv()` return Err — exits cleanly.
            let plain_env = std::env::var("KEBAB_PROGRESS")
-                .map(|v| v.eq_ignore_ascii_case("plain"))
-                .unwrap_or(false);
+                .is_ok_and(|v| v.eq_ignore_ascii_case("plain"));
            let mode = progress::ProgressMode::from_flags(cli.json, cli.quiet, plain_env);

            // Surface the active embedding backend/device on the terminal so the
@@ -828,6 +858,8 @@ fn run(cli: &Cli) -> anyhow::Result<()> {
            doc_id,
            repo,
            code_lang,
+            source_type,
+            source,
            trace,
            bulk,
        } => {
@@ -967,6 +999,8 @@ fn run(cli: &Cli) -> anyhow::Result<()> {
                doc_id: doc_id.as_ref().map(|s| kebab_core::DocumentId(s.clone())),
                repo: repo.clone(),
                code_lang: code_lang.clone(),
+                source_type: source_type.clone(),
+                source_id: source.clone(),
            };

            let q = kebab_core::SearchQuery {
--- a/crates/kebab-config/src/lib.rs
+++ b/crates/kebab-config/src/lib.rs
@@ -12,6 +12,12 @@ mod paths;
 pub mod migrate;
 pub use paths::{expand_path, expand_path_with_base};

+/// Implicit source id used when a single-root `[workspace]` config (no
+/// `[[workspace.sources]]`) is normalized into the multi-source model, and
+/// the `DEFAULT` value of the `documents.source_id` column. Kept in sync
+/// with the migration default in `migrations/V0XX__documents_source_id.sql`.
+pub const DEFAULT_SOURCE_ID: &str = "default";
+
 /// f32 의 shortest round-trip(Display)을 f64 로 재파싱해 직렬화한다.
 /// `0.3_f32` 가 `0.30000001192092896` 으로 새지 않고 `0.3` 으로 출력되게 한다.
 /// 마이그레이션 시 toml_edit relocation 의 무손실 비교를 깨지 않도록, 그리고
@@ -88,8 +94,67 @@ pub struct Config {

 #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
 pub struct WorkspaceCfg {
-    pub root: String,
+    /// Single-root workspace (legacy / common case). `Option` so that a
+    /// config that declares only `[[workspace.sources]]` (no bare `root`)
+    /// parses — and, symmetrically, a legacy single-`root` config (no
+    /// `sources`) still parses unchanged. The load-time normalizer
+    /// ([`Config::normalize_sources`]) reconciles the two into a single
+    /// non-empty `sources` list (`id = "default"` synthesized from `root`).
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub root: Option<String>,
    pub exclude: Vec<String>,
+    /// `[[workspace.sources]]`: named multi-source declaration. When empty
+    /// and `root` is set, the load path normalizes to a single implicit
+    /// `default` source. Each entry stamps its `id` onto every document it
+    /// ingests and supplies per-source `trust_level` / `source_type`
+    /// defaults (frontmatter still wins per the §0 Q9 derive table).
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub sources: Vec<SourceCfg>,
+}
+
+/// One named source under `[[workspace.sources]]`.
+///
+/// `trust_level` / `source_type` are the **source-level defaults**: they
+/// apply when a document's frontmatter does not specify the field. The
+/// precedence is `frontmatter > source default > hardcoded`
+/// (`TrustLevel::Primary` / `SourceType::Markdown`) — implemented in the
+/// markdown derive via `BodyHints::fallback_trust_level`.
+#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
+pub struct SourceCfg {
+    /// Stable identifier stamped onto `documents.source_id` for every
+    /// document ingested from this source. Must be unique and non-empty
+    /// across the workspace (enforced in [`Config::validate`]).
+    pub id: String,
+    /// Root directory to walk for this source. Accepts the same
+    /// absolute / `~` / `${VAR}` / relative(=config-dir-based) forms as
+    /// the legacy `workspace.root`.
+    pub root: String,
+    /// Per-source denylist globs, merged on top of `workspace.exclude`.
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub exclude: Vec<String>,
+    /// Per-source default `trust_level` (frontmatter overrides it).
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub trust_level: Option<kebab_core::TrustLevel>,
+    /// Per-source default `source_type` (frontmatter overrides it).
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub source_type: Option<kebab_core::SourceType>,
+}
+
+/// A source with its `root` resolved to an absolute path and its `exclude`
+/// merged with `workspace.exclude`. Produced by [`Config::resolved_sources`]
+/// — the single entry point the ingest pipeline iterates over.
+#[derive(Clone, Debug, PartialEq)]
+pub struct ResolvedSource {
+    /// Stamped onto `documents.source_id`.
+    pub id: String,
+    /// Absolute walk root (tilde / `${VAR}` / relative-to-config resolved).
+    pub root: PathBuf,
+    /// `workspace.exclude` ∪ per-source `exclude`.
+    pub exclude: Vec<String>,
+    /// Per-source default trust level (None → fall back to `Primary`).
+    pub trust_level: Option<kebab_core::TrustLevel>,
+    /// Per-source default source type (None → fall back to `Markdown`).
+    pub source_type: Option<kebab_core::SourceType>,
 }

 #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
@@ -782,12 +847,13 @@ impl Config {
        Self {
            schema_version: crate::migrate::CURRENT_SCHEMA_VERSION,
            workspace: WorkspaceCfg {
-                root: "~/KnowledgeBase".to_string(),
+                root: Some("~/KnowledgeBase".to_string()),
                exclude: vec![
                    ".git/**".to_string(),
                    "node_modules/**".to_string(),
                    ".obsidian/**".to_string(),
                ],
+                sources: vec![],
            },
            storage: StorageCfg {
                data_dir: "${XDG_DATA_HOME:-~/.local/share}/kebab".to_string(),
@@ -906,7 +972,78 @@ impl Config {
                PathBuf::from(".")
            })
        });
-        paths::expand_path_with_base(&self.workspace.root, "", &base)
+        paths::expand_path_with_base(&self.primary_root_raw(), "", &base)
+    }
+
+    /// The raw (unexpanded) string for the *primary* workspace root, used by
+    /// [`resolve_workspace_root`](Self::resolve_workspace_root) and any
+    /// single-root code path. Order: first `[[workspace.sources]]` entry's
+    /// `root` → bare `workspace.root` → `~/KnowledgeBase` default. This keeps
+    /// every pre-existing single-root call site working when only `sources`
+    /// is declared.
+    fn primary_root_raw(&self) -> String {
+        if let Some(s) = self.workspace.sources.first() {
+            return s.root.clone();
+        }
+        self.workspace
+            .root
+            .clone()
+            .unwrap_or_else(|| "~/KnowledgeBase".to_string())
+    }
+
+    /// The base directory for resolving relative source roots: the config
+    /// file's directory when loaded from disk, else the current dir (mirrors
+    /// [`resolve_workspace_root`](Self::resolve_workspace_root)).
+    fn root_resolution_base(&self) -> PathBuf {
+        self.source_dir.clone().unwrap_or_else(|| {
+            std::env::current_dir().unwrap_or_else(|e| {
+                tracing::warn!(
+                    target: "kebab-config",
+                    error = %e,
+                    "current_dir() failed; falling back to '.' for source root resolution"
+                );
+                PathBuf::from(".")
+            })
+        })
+    }
+
+    /// Normalized, resolved list of sources to ingest. Always non-empty:
+    ///
+    /// - If `[[workspace.sources]]` is declared, each entry is returned with
+    ///   its `root` expanded and `exclude` merged with `workspace.exclude`.
+    /// - Otherwise a single implicit source `id = "default"` is synthesized
+    ///   from `workspace.root` (the legacy single-root path).
+    ///
+    /// This is the single entry point the ingest pipeline iterates over, so
+    /// single-root and multi-source configs share one code path.
+    pub fn resolved_sources(&self) -> Vec<ResolvedSource> {
+        let base = self.root_resolution_base();
+        if self.workspace.sources.is_empty() {
+            let root = paths::expand_path_with_base(&self.primary_root_raw(), "", &base);
+            return vec![ResolvedSource {
+                id: DEFAULT_SOURCE_ID.to_string(),
+                root,
+                exclude: self.workspace.exclude.clone(),
+                trust_level: None,
+                source_type: None,
+            }];
+        }
+        self.workspace
+            .sources
+            .iter()
+            .map(|s| {
+                let root = paths::expand_path_with_base(&s.root, "", &base);
+                let mut exclude = self.workspace.exclude.clone();
+                exclude.extend(s.exclude.iter().cloned());
+                ResolvedSource {
+                    id: s.id.clone(),
+                    root,
+                    exclude,
+                    trust_level: s.trust_level,
+                    source_type: s.source_type,
+                }
+            })
+            .collect()
    }

    /// Read config from disk and merge env overrides on top of it. If the
@@ -1019,10 +1156,41 @@ impl Config {
                cause: format!("parse_failed: {e}"),
            })
        })?;
+        cfg.validate_sources().map_err(|cause| {
+            anyhow::Error::new(ConfigInvalid {
+                path: path.to_path_buf(),
+                cause,
+            })
+        })?;
        cfg.source_dir = path.parent().map(Path::to_path_buf);
        Ok(cfg)
    }

+    /// Validate `[[workspace.sources]]`: every `id` must be non-empty and
+    /// unique across the workspace. Empty `sources` (legacy single-root) is
+    /// always valid. Returns the failure cause string for `ConfigInvalid`.
+    fn validate_sources(&self) -> Result<(), String> {
+        let mut seen = std::collections::HashSet::new();
+        for s in &self.workspace.sources {
+            if s.id.trim().is_empty() {
+                return Err("workspace.sources: an entry has an empty `id`".to_string());
+            }
+            if s.root.trim().is_empty() {
+                return Err(format!(
+                    "workspace.sources: source `{}` has an empty `root`",
+                    s.id
+                ));
+            }
+            if !seen.insert(s.id.as_str()) {
+                return Err(format!(
+                    "workspace.sources: duplicate source id `{}` (ids must be unique)",
+                    s.id
+                ));
+            }
+        }
+        Ok(())
+    }
+
    /// Apply `KEBAB_<SECTION>_<KEY>` env overrides. Unknown keys are ignored.
    ///
    /// The mapping is an explicit grep-friendly whitelist — one match arm
@@ -1037,7 +1205,7 @@ impl Config {
            }
            match k.as_str() {
                // workspace
-                "KEBAB_WORKSPACE_ROOT" => self.workspace.root = v.clone(),
+                "KEBAB_WORKSPACE_ROOT" => self.workspace.root = Some(v.clone()),

                // storage
                "KEBAB_STORAGE_DATA_DIR" => self.storage.data_dir = v.clone(),
@@ -2034,7 +2202,7 @@ max_context_tokens = 8000
    #[test]
    fn legacy_include_field_is_ignored_silently() {
        let mut cfg = Config::defaults();
-        cfg.workspace.root = "/tmp/kebab-legacy".to_string();
+        cfg.workspace.root = Some("/tmp/kebab-legacy".to_string());
        let mut toml_text = toml::to_string(&cfg).expect("default round-trips");
        // Inject a legacy `include = [...]` line into the [workspace] block.
        toml_text = toml_text.replace(
@@ -2048,20 +2216,105 @@ max_context_tokens = 8000
            parsed.err()
        );
        let cfg = parsed.unwrap();
-        assert_eq!(cfg.workspace.root, "/tmp/kebab-legacy");
+        assert_eq!(cfg.workspace.root.as_deref(), Some("/tmp/kebab-legacy"));
    }

    /// p9-fb-25: `WorkspaceCfg` must NOT have an `include` field.
    /// Compile-time proof: exhaustive destructure.
    #[test]
-    fn workspace_cfg_has_only_root_and_exclude_fields() {
+    fn workspace_cfg_has_only_root_exclude_sources_fields() {
        let ws = Config::defaults().workspace;
        let WorkspaceCfg {
            root: _,
            exclude: _,
+            sources: _,
        } = &ws;
    }

+    #[test]
+    fn legacy_single_root_normalizes_to_default_source() {
+        // A single-root config (no [[workspace.sources]]) must resolve to
+        // exactly one source `id = "default"` rooted at workspace.root.
+        let mut cfg = Config::defaults();
+        cfg.workspace.root = Some("/tmp/kb-notes".to_string());
+        let resolved = cfg.resolved_sources();
+        assert_eq!(resolved.len(), 1);
+        assert_eq!(resolved[0].id, DEFAULT_SOURCE_ID);
+        assert_eq!(resolved[0].root, std::path::PathBuf::from("/tmp/kb-notes"));
+        assert_eq!(resolved[0].trust_level, None);
+    }
+
+    #[test]
+    fn multi_source_config_resolves_each_with_merged_exclude() {
+        let mut cfg = Config::defaults();
+        cfg.workspace.root = None;
+        cfg.workspace.exclude = vec![".git/**".to_string()];
+        cfg.workspace.sources = vec![
+            SourceCfg {
+                id: "notes".to_string(),
+                root: "/tmp/notes".to_string(),
+                exclude: vec![],
+                trust_level: Some(kebab_core::TrustLevel::Primary),
+                source_type: None,
+            },
+            SourceCfg {
+                id: "refs".to_string(),
+                root: "/tmp/refs".to_string(),
+                exclude: vec!["draft/**".to_string()],
+                trust_level: Some(kebab_core::TrustLevel::Secondary),
+                source_type: Some(kebab_core::SourceType::Reference),
+            },
+        ];
+        // A multi-source config (no bare root) must round-trip through TOML.
+        let toml_text = toml::to_string(&cfg).expect("multi-source serializes");
+        let cfg: Config = toml::from_str(&toml_text).expect("multi-source parses");
+        cfg.validate_sources().expect("valid sources");
+        let resolved = cfg.resolved_sources();
+        assert_eq!(resolved.len(), 2);
+        assert_eq!(resolved[0].id, "notes");
+        assert_eq!(resolved[0].root, std::path::PathBuf::from("/tmp/notes"));
+        assert_eq!(resolved[0].exclude, vec![".git/**".to_string()]);
+        assert_eq!(resolved[0].trust_level, Some(kebab_core::TrustLevel::Primary));
+        assert_eq!(resolved[1].id, "refs");
+        // workspace.exclude ∪ per-source exclude.
+        assert_eq!(
+            resolved[1].exclude,
+            vec![".git/**".to_string(), "draft/**".to_string()]
+        );
+        assert_eq!(
+            resolved[1].source_type,
+            Some(kebab_core::SourceType::Reference)
+        );
+        assert_eq!(
+            resolved[1].trust_level,
+            Some(kebab_core::TrustLevel::Secondary)
+        );
+    }
+
+    fn source_cfg(id: &str, root: &str) -> SourceCfg {
+        SourceCfg {
+            id: id.to_string(),
+            root: root.to_string(),
+            exclude: vec![],
+            trust_level: None,
+            source_type: None,
+        }
+    }
+
+    #[test]
+    fn duplicate_source_ids_rejected() {
+        let mut cfg = Config::defaults();
+        cfg.workspace.sources = vec![source_cfg("dup", "/a"), source_cfg("dup", "/b")];
+        assert!(cfg.validate_sources().is_err(), "duplicate ids must fail");
+    }
+
+    #[test]
+    fn empty_source_id_rejected() {
+        let mut cfg = Config::defaults();
+        cfg.workspace.sources = vec![source_cfg("", "/a")];
+        assert!(cfg.validate_sources().is_err(), "empty id must fail");
+    }
+
    #[test]
    fn default_stale_threshold_is_30() {
        let c = Config::defaults();
--- a/crates/kebab-config/src/migrate.rs
+++ b/crates/kebab-config/src/migrate.rs
@@ -9,7 +9,7 @@ use toml_edit::{DocumentMut, Item};

 /// 현재 바이너리가 이해하는 config 스키마 버전. 마이그레이션 완료 시
 /// 사용자 파일의 `schema_version` 을 이 값으로 stamp 한다.
-pub const CURRENT_SCHEMA_VERSION: u32 = 3;
+pub const CURRENT_SCHEMA_VERSION: u32 = 4;

 /// 한 번의 마이그레이션에서 발생한 개별 변경.
 #[derive(Clone, Debug, PartialEq, serde::Serialize)]
@@ -68,6 +68,7 @@ const HEADER: &str = "\
 fn section_comment(path: &str) -> Option<&'static str> {
    Some(match path {
        "workspace" => "# 색인 대상 워크스페이스.",
+        "workspace.sources" => "# named multi-source (각 source 의 id 가 documents.source_id 로 stamp).",
        "storage" => "# XDG 저장 경로(데이터/sqlite/벡터/에셋/모델).",
        "indexing" => "# 병렬도 + 파일시스템 watch.",
        "chunking" => "# 청크 크기·오버랩·heading 존중.",
@@ -376,6 +377,39 @@ pub fn step_2_to_3(doc: &mut DocumentMut, changes: &mut Vec<MigrationChange>) {
    copy_image_paddle_to_pdf(doc);
 }

+/// v3 → v4: 단일 `workspace.root` 를 `[[workspace.sources]]` 의 implicit
+/// `default` source 로 미러링한다(`id = "default"`, `root = <기존 root>`).
+/// 기존 `workspace.root` 키는 그대로 둔다 — `resolved_sources()` 가 sources
+/// 가 있으면 그쪽을 우선하므로 무해하고, defaults reconcile 이 root 를 다시
+/// 추가하려 하지 않게 한다. 멱등: `[[workspace.sources]]` 가 이미 있으면 no-op.
+pub fn step_3_to_4(doc: &mut DocumentMut, changes: &mut Vec<MigrationChange>) {
+    let Some(ws) = doc.get_mut("workspace").and_then(Item::as_table_mut) else {
+        return;
+    };
+    // 이미 sources 가 선언돼 있으면(array-of-tables 든 inline 이든) 손대지 않음.
+    if ws.contains_key("sources") {
+        return;
+    }
+    // root 가 없으면 만들 게 없음(defaults 에는 항상 있지만 방어).
+    let Some(root_val) = ws.get("root").and_then(Item::as_str).map(str::to_string) else {
+        return;
+    };
+
+    let mut entry = toml_edit::Table::new();
+    entry.insert("id", toml_edit::value("default"));
+    entry.insert("root", toml_edit::value(root_val));
+
+    let mut aot = toml_edit::ArrayOfTables::new();
+    aot.push(entry);
+    ws.insert("sources", Item::ArrayOfTables(aot));
+
+    changes.push(MigrationChange {
+        kind: ChangeKind::AddedSection,
+        path: "workspace.sources".to_string(),
+        detail: "workspace.root → [[workspace.sources]] id=default".to_string(),
+    });
+}
+
 /// 파일의 schema_version(없으면 1) 부터 CURRENT 까지 step 적용.
 fn run_steps(doc: &mut DocumentMut, from: u32, changes: &mut Vec<MigrationChange>) {
    if from < 2 {
@@ -384,6 +418,9 @@ fn run_steps(doc: &mut DocumentMut, from: u32, changes: &mut Vec<MigrationChange
    if from < 3 {
        step_2_to_3(doc, changes);
    }
+    if from < 4 {
+        step_3_to_4(doc, changes);
+    }
 }

 /// 사용자 config.toml 텍스트를 받아 step 체인 + reconciliation + version
@@ -648,6 +685,76 @@ engine = \"paddle-onnx\"
        assert!(again.is_empty(), "not idempotent: {again:?}");
    }

+    #[test]
+    fn step_3_to_4_mirrors_root_into_default_source() {
+        let v3 = "\
+schema_version = 3
+
+[workspace]
+root = \"/my/notes\"
+exclude = [\".git/**\"]
+";
+        let mut doc: DocumentMut = v3.parse().unwrap();
+        let mut changes = Vec::new();
+        step_3_to_4(&mut doc, &mut changes);
+        let out = doc.to_string();
+        // 새 array-of-tables 가 id=default 로 추가.
+        assert!(out.contains("[[workspace.sources]]"), "{out}");
+        assert!(out.contains("id = \"default\""), "{out}");
+        // 기존 root 는 보존(reconcile 이 다시 추가하지 않게).
+        assert!(out.contains("root = \"/my/notes\""), "{out}");
+        // 재파싱 후 sources.default 가 root 를 미러.
+        let reparsed: DocumentMut = out.parse().unwrap();
+        let src0 = reparsed["workspace"]["sources"][0].as_table().unwrap();
+        assert_eq!(src0["id"].as_str(), Some("default"));
+        assert_eq!(src0["root"].as_str(), Some("/my/notes"));
+        // 멱등.
+        let mut changes2 = Vec::new();
+        step_3_to_4(&mut doc, &mut changes2);
+        assert!(changes2.is_empty(), "step_3_to_4 not idempotent");
+    }
+
+    #[test]
+    fn step_3_to_4_noop_when_sources_already_present() {
+        let v4 = "\
+schema_version = 4
+
+[workspace]
+root = \"/my/notes\"
+exclude = []
+
+[[workspace.sources]]
+id = \"notes\"
+root = \"/my/notes\"
+";
+        let mut doc: DocumentMut = v4.parse().unwrap();
+        let mut changes = Vec::new();
+        step_3_to_4(&mut doc, &mut changes);
+        assert!(changes.is_empty(), "must not touch existing sources");
+        // 기존 source 만 존재(default 가 추가되지 않음).
+        assert!(!doc.to_string().contains("id = \"default\""));
+    }
+
+    #[test]
+    fn migrate_document_v3_to_v4_adds_sources_and_is_idempotent() {
+        let v3 = "\
+schema_version = 3
+
+[workspace]
+root = \"/n\"
+exclude = []
+";
+        let outcome = migrate_document(v3);
+        assert_eq!(outcome.from_schema_version, 3);
+        assert_eq!(outcome.to_schema_version, 4);
+        assert!(outcome.changed());
+        assert!(outcome.new_text.contains("[[workspace.sources]]"));
+        assert_eq!(read_schema_version(&outcome.new_text), 4);
+        let again = migrate_document(&outcome.new_text);
+        assert!(!again.changed(), "not idempotent: {:?}", again.changes);
+        assert_eq!(again.new_text, outcome.new_text);
+    }
+
    #[test]
    fn migrate_document_missing_schema_version_treated_as_v1() {
        let old = "[workspace]\nroot = \"/n\"\n";
--- a/crates/kebab-config/tests/migrate_v3.rs
+++ b/crates/kebab-config/tests/migrate_v3.rs
@@ -11,11 +11,16 @@ const USER_V2: &str = include_str!("fixtures/user_v2_config.toml");
 fn user_v2_migrates_losslessly() {
    let out = migrate_document(USER_V2);
    assert_eq!(out.from_schema_version, 2);
-    assert_eq!(out.to_schema_version, 3);
+    // v2 → CURRENT(=4): v3 의 [ingest.*] relocation 에 더해 v4 의
+    // [[workspace.sources]] default source 미러링까지 적용된다.
+    assert_eq!(out.to_schema_version, 4);
    let t = &out.new_text;

    // 사용자 값 보존.
    assert!(t.contains("root = \"/Users/user/Obsidian/Default\""), "{t}");
+    // v4: workspace.root → [[workspace.sources]] id=default 미러링.
+    assert!(t.contains("[[workspace.sources]]"), "v4 sources 누락:\n{t}");
+    assert!(t.contains("id = \"default\""), "default source 누락:\n{t}");
    assert!(t.contains("model = \"snowflake-arctic-embed2\""));
    assert!(t.contains("endpoint = \"http://192.168.0.2:11943\""));
    // 사용자 주석/대안 줄 보존.
--- a/crates/kebab-core/src/metadata.rs
+++ b/crates/kebab-core/src/metadata.rs
@@ -36,6 +36,14 @@ pub struct Metadata {
    /// for markdown / pdf / image. Set by the local-filesystem source connector during ingest.
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub code_lang: Option<String>,
+
+    /// `[[workspace.sources]]`: id of the named source this document was
+    /// ingested from (the `id` of the matching `[[workspace.sources]]`
+    /// entry; `"default"` for single-root workspaces normalized to the
+    /// implicit `default` source). null on documents ingested before the
+    /// multi-source feature; the store column defaults to `"default"`.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub source_id: Option<String>,
 }

 #[derive(Clone, Copy, Debug, Eq, Hash, PartialEq, Serialize, Deserialize)]
@@ -105,12 +113,14 @@ mod tests {
            git_branch: None,
            git_commit: None,
            code_lang: None,
+            source_id: None,
        };
        let v = serde_json::to_value(&m).unwrap();
        assert!(v.get("repo").is_none());
        assert!(v.get("git_branch").is_none());
        assert!(v.get("git_commit").is_none());
        assert!(v.get("code_lang").is_none());
+        assert!(v.get("source_id").is_none());
    }

    #[test]
@@ -128,8 +138,10 @@ mod tests {
            git_branch: Some("main".into()),
            git_commit: Some("a".repeat(40)),
            code_lang: Some("rust".into()),
+            source_id: Some("notes".into()),
        };
        let v = serde_json::to_value(&m).unwrap();
+        assert_eq!(v["source_id"], "notes");
        assert_eq!(v["repo"], "kebab");
        assert_eq!(v["git_branch"], "main");
        assert_eq!(v["git_commit"].as_str().unwrap().len(), 40);
--- a/crates/kebab-core/src/search.rs
+++ b/crates/kebab-core/src/search.rs
@@ -69,6 +69,20 @@ pub struct SearchFilters {
    /// Unknown values produce empty hits (consistent with `media` policy).
    #[serde(default)]
    pub code_lang: Vec<String>,
+    /// Phase-2 (jira-contamination experiment): filter by `documents.source_type`
+    /// (`markdown` | `note` | `paper` | `reference` | `inbox`). Empty = no filter;
+    /// multi-value = OR. Direct indexed column — the clean provenance/source lever:
+    /// filtering recovers concept-query precision without the see-saw of global
+    /// trust-weighting (see tasks/HOTFIXES.md A/B evidence).
+    #[serde(default)]
+    pub source_type: Vec<String>,
+    /// `[[workspace.sources]]`: filter by `documents.source_id` (the `id` of
+    /// the `[[workspace.sources]]` entry a document was ingested from; e.g.
+    /// `default`, `notes`, `code`). Empty = no filter; multi-value = OR.
+    /// Direct indexed column (idx_docs_source_id) — the named-source
+    /// provenance lever for multi-source KBs.
+    #[serde(default)]
+    pub source_id: Vec<String>,
 }

 #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
--- a/crates/kebab-mcp/src/tools/search.rs
+++ b/crates/kebab-mcp/src/tools/search.rs
@@ -107,6 +107,8 @@ pub fn handle(state: &KebabAppState, input: SearchInput) -> CallToolResult {
        doc_id: input.doc_id.clone().map(kebab_core::DocumentId),
        repo: vec![],
        code_lang: vec![],
+        source_type: vec![],
+        source_id: vec![],
    };

    let query = kebab_core::SearchQuery {
--- a/crates/kebab-mcp/tests/error_mapping.rs
+++ b/crates/kebab-mcp/tests/error_mapping.rs
@@ -10,7 +10,7 @@ async fn schema_tool_emits_error_v1_when_db_missing() {
    let dir = tempfile::tempdir().unwrap();
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = dir.path().to_string_lossy().into_owned();
-    cfg.workspace.root = dir.path().join("notes").to_string_lossy().into_owned();
+    cfg.workspace.root = Some(dir.path().join("notes").to_string_lossy().into_owned());
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
    // Note: NO ingest call — kebab.sqlite is absent → schema_with_config
--- a/crates/kebab-mcp/tests/tools_call_ask.rs
+++ b/crates/kebab-mcp/tests/tools_call_ask.rs
@@ -10,7 +10,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_ask_multi_hop.rs
+++ b/crates/kebab-mcp/tests/tools_call_ask_multi_hop.rs
@@ -27,7 +27,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_bulk_search.rs
+++ b/crates/kebab-mcp/tests/tools_call_bulk_search.rs
@@ -12,7 +12,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_doctor.rs
+++ b/crates/kebab-mcp/tests/tools_call_doctor.rs
@@ -9,10 +9,10 @@ async fn doctor_tool_returns_doctor_v1_json() {
    let dir = tempfile::tempdir().unwrap();
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = dir.path().join("data").to_string_lossy().into_owned();
-    cfg.workspace.root = dir.path().join("notes").to_string_lossy().into_owned();
+    cfg.workspace.root = Some(dir.path().join("notes").to_string_lossy().into_owned());
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
-    std::fs::create_dir_all(&cfg.workspace.root).unwrap();
+    std::fs::create_dir_all(cfg.resolve_workspace_root()).unwrap();

    // Pass None for config_path — doctor falls back to XDG default probe
    // (path won't exist in the tempdir, which is fine; doctor reports it
--- a/crates/kebab-mcp/tests/tools_call_fetch.rs
+++ b/crates/kebab-mcp/tests/tools_call_fetch.rs
@@ -16,7 +16,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_ingest_file.rs
+++ b/crates/kebab-mcp/tests/tools_call_ingest_file.rs
@@ -15,7 +15,7 @@ async fn ingest_file_tool_returns_ingest_report_v1() {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
@@ -61,7 +61,7 @@ async fn ingest_file_tool_idempotent_on_second_call() {
    std::fs::create_dir_all(&data).unwrap();

    let mut cfg = kebab_config::Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_ingest_stdin.rs
+++ b/crates/kebab-mcp/tests/tools_call_ingest_stdin.rs
@@ -14,7 +14,7 @@ fn fresh_state(dir: &std::path::Path) -> KebabAppState {
    fs::create_dir_all(&data).unwrap();

    let mut cfg = Config::defaults();
-    cfg.workspace.root = workspace.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace.to_string_lossy().into_owned());
    cfg.storage.data_dir = data.to_string_lossy().into_owned();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_schema.rs
+++ b/crates/kebab-mcp/tests/tools_call_schema.rs
@@ -11,7 +11,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_search.rs
+++ b/crates/kebab-mcp/tests/tools_call_search.rs
@@ -11,7 +11,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-mcp/tests/tools_call_search_trace.rs
+++ b/crates/kebab-mcp/tests/tools_call_search_trace.rs
@@ -11,7 +11,7 @@ fn minimal_config(data_dir: &std::path::Path, workspace_root: &std::path::Path)
    let mut cfg = Config::defaults();
    cfg.storage.data_dir = data_dir.to_string_lossy().into_owned();
    cfg.storage.model_dir = data_dir.join("models").to_string_lossy().into_owned();
-    cfg.workspace.root = workspace_root.to_string_lossy().into_owned();
+    cfg.workspace.root = Some(workspace_root.to_string_lossy().into_owned());
    cfg.workspace.exclude.clear();
    cfg.models.embedding.provider = "none".to_string();
    cfg.models.embedding.dimensions = 0;
--- a/crates/kebab-parse-code/src/c.rs
+++ b/crates/kebab-parse-code/src/c.rs
@@ -131,6 +131,7 @@ impl Extractor for CAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("c".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/cpp.rs
+++ b/crates/kebab-parse-code/src/cpp.rs
@@ -155,6 +155,7 @@ impl Extractor for CppAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("cpp".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/go.rs
+++ b/crates/kebab-parse-code/src/go.rs
@@ -133,6 +133,7 @@ impl Extractor for GoAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("go".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/java.rs
+++ b/crates/kebab-parse-code/src/java.rs
@@ -144,6 +144,7 @@ impl Extractor for JavaAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("java".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/javascript.rs
+++ b/crates/kebab-parse-code/src/javascript.rs
@@ -151,6 +151,7 @@ impl Extractor for JavascriptAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("javascript".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/kotlin.rs
+++ b/crates/kebab-parse-code/src/kotlin.rs
@@ -149,6 +149,7 @@ impl Extractor for KotlinAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("kotlin".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/python.rs
+++ b/crates/kebab-parse-code/src/python.rs
@@ -133,6 +133,7 @@ impl Extractor for PythonAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("python".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/rust.rs
+++ b/crates/kebab-parse-code/src/rust.rs
@@ -136,6 +136,7 @@ impl Extractor for RustAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("rust".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-code/src/typescript.rs
+++ b/crates/kebab-parse-code/src/typescript.rs
@@ -144,6 +144,7 @@ impl Extractor for TypescriptAstExtractor {
            git_branch,
            git_commit,
            code_lang: Some("typescript".to_string()),
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-image/src/lib.rs
+++ b/crates/kebab-parse-image/src/lib.rs
@@ -203,6 +203,7 @@ impl Extractor for ImageExtractor {
            git_branch: None,
            git_commit: None,
            code_lang: None,
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-parse-md/src/frontmatter.rs
+++ b/crates/kebab-parse-md/src/frontmatter.rs
@@ -42,6 +42,16 @@ pub struct BodyHints {
    /// Optional language fallback used when neither frontmatter nor lingua
    /// detection produce a value. If `None` the final fallback is `"und"`.
    pub fallback_lang: Option<String>,
+    /// `[[workspace.sources]]`: id of the source this document is being
+    /// ingested from. Copied verbatim into `Metadata.source_id` (frontmatter
+    /// does not override the source id — it is an ingest-time provenance
+    /// stamp, not a user-authored field). `None` when single-root /
+    /// unspecified, in which case `Metadata.source_id` stays `None`.
+    pub source_id: Option<String>,
+    /// `[[workspace.sources]]`: per-source default `trust_level`. Consulted
+    /// only when the frontmatter does not specify `trust_level`. Precedence:
+    /// frontmatter > this source default > hardcoded `Primary`.
+    pub fallback_trust_level: Option<TrustLevel>,
 }

 /// Byte range of the frontmatter region inside the input slice.
@@ -444,8 +454,12 @@ fn derive_metadata(
    };

    // ---- trust_level ----
+    // Precedence: frontmatter > per-source default (hints.fallback_trust_level)
+    // > hardcoded Primary. An *unknown* frontmatter value warns and also falls
+    // through to the source default (then Primary), so a typo doesn't silently
+    // promote past the source's intended trust.
    let trust_level = match raw.trust_level.as_deref() {
-        None => TrustLevel::Primary,
+        None => hints.fallback_trust_level.unwrap_or(TrustLevel::Primary),
        Some(s) => {
            if let Some(tl) = parse_trust_level(s) {
                tl
@@ -454,7 +468,7 @@ fn derive_metadata(
                    kind: WarningKind::MalformedFrontmatter,
                    note: format!("unknown trust_level={s}, defaulted to primary"),
                });
-                TrustLevel::Primary
+                hints.fallback_trust_level.unwrap_or(TrustLevel::Primary)
            }
        }
    };
@@ -477,6 +491,10 @@ fn derive_metadata(
        git_branch: None,
        git_commit: None,
        code_lang: None,
+        // `[[workspace.sources]]`: ingest-time provenance stamp. Frontmatter
+        // does not override the source id — it is supplied by the caller
+        // (kebab-app) from the matching source's config `id`.
+        source_id: hints.source_id.clone(),
    }
 }

@@ -604,6 +622,8 @@ mod tests {
            fs_ctime: datetime!(2024-01-01 00:00:00 UTC),
            fs_mtime: datetime!(2024-01-02 00:00:00 UTC),
            fallback_lang: None,
+            source_id: None,
+            fallback_trust_level: None,
        }
    }

@@ -695,6 +715,47 @@ source_type: alien\n\
        assert!(warns.iter().any(|w| w.note.contains("source_type=alien")));
    }

+    fn hints_with_source(id: &str, trust: Option<TrustLevel>) -> BodyHints {
+        BodyHints {
+            source_id: Some(id.to_string()),
+            fallback_trust_level: trust,
+            ..hints()
+        }
+    }
+
+    #[test]
+    fn source_default_trust_applied_when_frontmatter_absent() {
+        // No `trust_level:` in frontmatter → the per-source default wins
+        // over the hardcoded Primary.
+        let md = b"---\ntitle: Doc\n---\nbody\n";
+        let (meta, _span, warns) =
+            parse_frontmatter(md, &hints_with_source("notes", Some(TrustLevel::Secondary)))
+                .unwrap();
+        assert!(warns.is_empty(), "warnings: {warns:?}");
+        assert_eq!(meta.trust_level, TrustLevel::Secondary);
+        assert_eq!(meta.source_id.as_deref(), Some("notes"));
+    }
+
+    #[test]
+    fn frontmatter_trust_overrides_source_default() {
+        // Explicit frontmatter trust beats the per-source default.
+        let md = b"---\ntrust_level: generated\n---\nbody\n";
+        let (meta, _span, _warns) =
+            parse_frontmatter(md, &hints_with_source("notes", Some(TrustLevel::Secondary)))
+                .unwrap();
+        assert_eq!(meta.trust_level, TrustLevel::Generated);
+        assert_eq!(meta.source_id.as_deref(), Some("notes"));
+    }
+
+    #[test]
+    fn no_source_id_leaves_metadata_source_id_none() {
+        let md = b"---\ntitle: Doc\n---\nbody\n";
+        let (meta, _span, _warns) = parse_frontmatter(md, &hints()).unwrap();
+        assert_eq!(meta.source_id, None);
+        // Without a source default, hardcoded Primary still applies.
+        assert_eq!(meta.trust_level, TrustLevel::Primary);
+    }
+
    #[test]
    fn malformed_yaml_emits_warning_and_defaults() {
        // Unclosed quote → YAML parse fails.
--- a/crates/kebab-parse-md/src/normalize.rs
+++ b/crates/kebab-parse-md/src/normalize.rs
@@ -469,6 +469,7 @@ mod tests {
            git_branch: None,
            git_commit: None,
            code_lang: None,
+            source_id: None,
        }
    }

--- a/crates/kebab-parse-md/tests/frontmatter_snapshots.rs
+++ b/crates/kebab-parse-md/tests/frontmatter_snapshots.rs
@@ -37,6 +37,8 @@ fn pinned_hints() -> BodyHints {
        fs_ctime: datetime!(2024-01-01 00:00:00 UTC),
        fs_mtime: datetime!(2024-01-02 00:00:00 UTC),
        fallback_lang: None,
+        source_id: None,
+        fallback_trust_level: None,
    }
 }

--- a/crates/kebab-parse-md/tests/normalize_snapshot.rs
+++ b/crates/kebab-parse-md/tests/normalize_snapshot.rs
@@ -86,6 +86,8 @@ fn code_and_table_canonical_snapshot() {
        fs_ctime: asset.discovered_at,
        fs_mtime: asset.discovered_at,
        fallback_lang: Some("en".into()),
+        source_id: None,
+        fallback_trust_level: None,
    };
    let (metadata, fm_span, _fm_warns) =
        parse_frontmatter(&bytes, &hints).expect("frontmatter parses");
--- a/crates/kebab-parse-pdf/src/lib.rs
+++ b/crates/kebab-parse-pdf/src/lib.rs
@@ -203,6 +203,7 @@ impl Extractor for PdfTextExtractor {
            git_branch: None,
            git_commit: None,
            code_lang: None,
+            source_id: None,
        };

        tracing::debug!(
--- a/crates/kebab-search/src/lexical.rs
+++ b/crates/kebab-search/src/lexical.rs
@@ -419,6 +419,31 @@ fn run_query(
        }
    }

+    // Phase-2: source_type filter (IN-list on the direct `documents.source_type`
+    // column). Empty Vec = no filter; multi-value = OR. Mirrors filters.rs.
+    if !filters.source_type.is_empty() {
+        let placeholders = std::iter::repeat_n("?", filters.source_type.len())
+            .collect::<Vec<_>>()
+            .join(",");
+        sql.push_str(&format!(" AND d.source_type IN ({placeholders})"));
+        for st in &filters.source_type {
+            params.push(Box::new(st.clone()));
+        }
+    }
+
+    // [[workspace.sources]]: source_id filter (IN-list on the direct
+    // `documents.source_id` column). Empty Vec = no filter; multi-value = OR.
+    // Mirrors filters.rs.
+    if !filters.source_id.is_empty() {
+        let placeholders = std::iter::repeat_n("?", filters.source_id.len())
+            .collect::<Vec<_>>()
+            .join(",");
+        sql.push_str(&format!(" AND d.source_id IN ({placeholders})"));
+        for sid in &filters.source_id {
+            params.push(Box::new(sid.clone()));
+        }
+    }
+
    // p9-fb-36: ingested_after filter.
    // `documents.updated_at` is RFC3339 stored as TEXT (always UTC `Z` per
    // fb-32 ingest path), so lexicographic >= compare is correct — but only
--- a/crates/kebab-source-fs/src/connector.rs
+++ b/crates/kebab-source-fs/src/connector.rs
@@ -231,6 +231,47 @@ pub struct FsScanSkips {
    pub events: Vec<FsSkipEvent>,
 }

+impl FsScanSkips {
+    /// `[[workspace.sources]]`: fold another source's scan skips into `self`,
+    /// so a multi-source ingest reports aggregate counts. Counters add;
+    /// per-category sample vecs concatenate and re-cap at 5 (spec §5.5);
+    /// events concatenate.
+    pub fn merge(&mut self, other: FsScanSkips) {
+        self.skipped_gitignore = self.skipped_gitignore.saturating_add(other.skipped_gitignore);
+        self.skipped_kebabignore = self
+            .skipped_kebabignore
+            .saturating_add(other.skipped_kebabignore);
+        self.skipped_builtin_blacklist = self
+            .skipped_builtin_blacklist
+            .saturating_add(other.skipped_builtin_blacklist);
+        self.skipped_generated = self.skipped_generated.saturating_add(other.skipped_generated);
+        self.skipped_size_exceeded = self
+            .skipped_size_exceeded
+            .saturating_add(other.skipped_size_exceeded);
+
+        fn merge_samples(dst: &mut Vec<String>, src: Vec<String>) {
+            for s in src {
+                if dst.len() >= 5 {
+                    break;
+                }
+                dst.push(s);
+            }
+        }
+        merge_samples(&mut self.skip_examples.generated, other.skip_examples.generated);
+        merge_samples(
+            &mut self.skip_examples.size_exceeded,
+            other.skip_examples.size_exceeded,
+        );
+        merge_samples(
+            &mut self.skip_examples.builtin_blacklist,
+            other.skip_examples.builtin_blacklist,
+        );
+        merge_samples(&mut self.skip_examples.gitignore, other.skip_examples.gitignore);
+
+        self.events.extend(other.events);
+    }
+}
+
 /// A single per-file skip event for structured ingest log (v0.20.x).
 #[derive(Debug)]
 pub struct FsSkipEvent {
@@ -326,7 +367,7 @@ mod tests {

    fn cfg_with_root(root: &str) -> Config {
        let mut c = Config::defaults();
-        c.workspace.root = root.to_string();
+        c.workspace.root = Some(root.to_string());
        c.workspace.exclude.clear();
        c
    }
--- a/crates/kebab-source-fs/tests/include_allowlist.rs
+++ b/crates/kebab-source-fs/tests/include_allowlist.rs
@@ -20,7 +20,7 @@ use kebab_source_fs::FsSourceConnector;

 fn cfg_with_root(root: &str) -> Config {
    let mut c = Config::defaults();
-    c.workspace.root = root.to_string();
+    c.workspace.root = Some(root.to_string());
    c.workspace.exclude.clear();
    // Disable size / generated caps so small test files always pass.
    c.ingest.code.max_file_bytes = u64::MAX;
--- a/crates/kebab-source-fs/tests/snapshot_tree1.rs
+++ b/crates/kebab-source-fs/tests/snapshot_tree1.rs
@@ -50,7 +50,7 @@ fn baseline_path() -> PathBuf {

 fn cfg_for_fixture(root: &str) -> Config {
    let mut c = Config::defaults();
-    c.workspace.root = root.to_string();
+    c.workspace.root = Some(root.to_string());
    // Clear default excludes (`.git/**`, `node_modules/**`, `.obsidian/**`)
    // so the snapshot is purely a function of the fixture + .kebabignore +
    // baked-in default-excludes.
--- a/crates/kebab-source-fs/tests/symlink_cycle.rs
+++ b/crates/kebab-source-fs/tests/symlink_cycle.rs
@@ -23,7 +23,7 @@ use kebab_source_fs::FsSourceConnector;

 fn cfg_with_root(root: &str) -> Config {
    let mut c = Config::defaults();
-    c.workspace.root = root.to_string();
+    c.workspace.root = Some(root.to_string());
    c.workspace.exclude.clear();
    c
 }
--- a/crates/kebab-store-sqlite/src/documents.rs
+++ b/crates/kebab-store-sqlite/src/documents.rs
@@ -745,6 +745,14 @@ fn upsert_document(
    // `markdown` for the column).
    let source_type = source_type_label(&doc.metadata.source_type);
    let trust_level = trust_level_label(&doc.metadata.trust_level);
+    // `[[workspace.sources]]`: id of the source this doc came from. Falls back
+    // to the column default `"default"` for docs without an explicit source
+    // (single-root workspaces / pre-multi-source ingests).
+    let source_id = doc
+        .metadata
+        .source_id
+        .as_deref()
+        .unwrap_or(kebab_config::DEFAULT_SOURCE_ID);
    let created_at = doc
        .metadata
        .created_at
@@ -757,11 +765,11 @@ fn upsert_document(
    tx.execute(
        "INSERT INTO documents (
            doc_id, asset_id, workspace_path, title, lang,
-            source_type, trust_level, parser_version,
+            source_type, trust_level, source_id, parser_version,
            doc_version, schema_version, metadata_json,
            provenance_json, created_at, updated_at,
            last_chunker_version, last_embedding_version
-        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
        ON CONFLICT(doc_id) DO UPDATE SET
            asset_id              = excluded.asset_id,
            workspace_path        = excluded.workspace_path,
@@ -769,6 +777,7 @@ fn upsert_document(
            lang                  = excluded.lang,
            source_type           = excluded.source_type,
            trust_level           = excluded.trust_level,
+            source_id             = excluded.source_id,
            parser_version        = excluded.parser_version,
            -- doc_version: bump on update. excluded.doc_version is the
            -- caller's submitted value; we ignore it and add 1 to the
@@ -788,6 +797,7 @@ fn upsert_document(
            doc.lang.0,
            source_type,
            trust_level,
+            source_id,
            doc.parser_version.0,
            i64::from(doc.doc_version),
            i64::from(doc.schema_version),
--- a/crates/kebab-store-sqlite/src/filters.rs
+++ b/crates/kebab-store-sqlite/src/filters.rs
@@ -191,6 +191,31 @@ impl SqliteStore {
            }
        }

+        // Phase-2: source_type filter (IN-list on the direct `documents.source_type`
+        // column, idx_docs_source_type). Empty Vec = no filter; multi-value = OR.
+        if !filters.source_type.is_empty() {
+            let placeholders = std::iter::repeat_n("?", filters.source_type.len())
+                .collect::<Vec<_>>()
+                .join(",");
+            sql.push_str(&format!(" AND d.source_type IN ({placeholders})"));
+            for st in &filters.source_type {
+                bind.push(Box::new(st.clone()));
+            }
+        }
+
+        // [[workspace.sources]]: source_id filter (IN-list on the direct
+        // `documents.source_id` column, idx_docs_source_id). Empty Vec = no
+        // filter; multi-value = OR. Mirrors the source_type filter above.
+        if !filters.source_id.is_empty() {
+            let placeholders = std::iter::repeat_n("?", filters.source_id.len())
+                .collect::<Vec<_>>()
+                .join(",");
+            sql.push_str(&format!(" AND d.source_id IN ({placeholders})"));
+            for sid in &filters.source_id {
+                bind.push(Box::new(sid.clone()));
+            }
+        }
+
        // p9-fb-36: ingested_after filter.
        // `documents.updated_at` is RFC3339 TEXT (UTC `Z` per fb-32);
        // lexicographic >= compare is correct — but only when the filter
@@ -1000,6 +1025,121 @@ mod tests {
        );
    }

+    /// [[workspace.sources]]: the `source_id` filter keeps only chunks whose
+    /// owning document's `documents.source_id` column is in the IN-list.
+    #[test]
+    fn filter_chunks_source_id_keeps_matching_source() {
+        let tmp = TempDir::new().unwrap();
+        let store = open_store(&tmp);
+        let c1 = "11111111111111111111111111111111";
+        let c2 = "22222222222222222222222222222222";
+        let c3 = "33333333333333333333333333333333";
+        // Three docs, each with a distinct source_id column value.
+        seed_with_source_id(&store, c1, "d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1d1", "notes/a.md", "notes");
+        seed_with_source_id(&store, c2, "d2d2d2d2d2d2d2d2d2d2d2d2d2d2d2d2", "code/b.rs", "code");
+        seed_with_source_id(
+            &store,
+            c3,
+            "d3d3d3d3d3d3d3d3d3d3d3d3d3d3d3d3",
+            "x.md",
+            "default",
+        );
+
+        // Single value.
+        let f = SearchFilters {
+            source_id: vec!["notes".to_string()],
+            ..Default::default()
+        };
+        let out = store
+            .filter_chunks(&[cid(c1), cid(c2), cid(c3)], &f)
+            .unwrap();
+        assert_eq!(out, vec![cid(c1)], "only the `notes` source chunk survives");
+
+        // Multi-value OR.
+        let f = SearchFilters {
+            source_id: vec!["notes".to_string(), "code".to_string()],
+            ..Default::default()
+        };
+        let out = store
+            .filter_chunks(&[cid(c1), cid(c2), cid(c3)], &f)
+            .unwrap();
+        assert_eq!(out, vec![cid(c1), cid(c2)], "notes OR code survive");
+
+        // Empty filter = no filtering.
+        let f = SearchFilters::default();
+        let out = store
+            .filter_chunks(&[cid(c1), cid(c2), cid(c3)], &f)
+            .unwrap();
+        assert_eq!(out, vec![cid(c1), cid(c2), cid(c3)]);
+    }
+
+    /// Seed one committed doc + chunk + embedding with an explicit
+    /// `documents.source_id` column value (the DEFAULT is `'default'`).
+    fn seed_with_source_id(
+        store: &SqliteStore,
+        chunk_id: &str,
+        doc_id: &str,
+        workspace_path: &str,
+        source_id: &str,
+    ) {
+        let asset_id = format!("a{}", &doc_id[..31]);
+        {
+            let conn = store.lock_conn();
+            conn.execute(
+                "INSERT INTO assets (
+                    asset_id, source_uri, workspace_path, media_type, byte_len,
+                    checksum, storage_kind, storage_path, discovered_at
+                 ) VALUES (?, ?, ?, '\"markdown\"', 1, ?, 'reference', ?,
+                           '1970-01-01T00:00:00Z')",
+                params![
+                    asset_id,
+                    format!("file://{workspace_path}"),
+                    workspace_path,
+                    workspace_path,
+                    workspace_path,
+                ],
+            )
+            .unwrap();
+            conn.execute(
+                "INSERT INTO documents (
+                    doc_id, asset_id, workspace_path, title, lang, source_type,
+                    trust_level, source_id, parser_version, doc_version,
+                    schema_version, metadata_json, provenance_json,
+                    created_at, updated_at
+                 ) VALUES (?, ?, ?, NULL, 'en', 'markdown', 'primary', ?, 'v1',
+                           1, 1, '{}', '{}', '1970-01-01T00:00:00Z',
+                           '1970-01-01T00:00:00Z')",
+                params![doc_id, asset_id, workspace_path, source_id],
+            )
+            .unwrap();
+            conn.execute(
+                "INSERT INTO chunks (
+                    chunk_id, doc_id, text, heading_path_json, section_label,
+                    source_spans_json, token_estimate, chunker_version,
+                    policy_hash, block_ids_json, created_at
+                 ) VALUES (?, ?, 'hi', '[]', NULL, '[]', 1, 'v1', 'h', '[]',
+                           '1970-01-01T00:00:00Z')",
+                params![chunk_id, doc_id],
+            )
+            .unwrap();
+        }
+        let embed_row = EmbeddingRecordRow {
+            embedding_id: format!("e{}", &chunk_id[..31]),
+            chunk_id: chunk_id.to_string(),
+            model_id: "m".to_string(),
+            model_version: "v1".to_string(),
+            dimensions: 4,
+            lance_table: "t".to_string(),
+            created_at: OffsetDateTime::UNIX_EPOCH,
+        };
+        store
+            .put_embedding_records_pending(std::slice::from_ref(&embed_row))
+            .unwrap();
+        store
+            .mark_embedding_records_committed(std::slice::from_ref(&embed_row.embedding_id))
+            .unwrap();
+    }
+
    #[test]
    fn filter_chunks_ingested_after_non_utc_offset_compares_as_instant() {
        // Regression test for the non-UTC offset lex-compare bug.
--- a/crates/kebab-store-sqlite/src/stats_ext.rs
+++ b/crates/kebab-store-sqlite/src/stats_ext.rs
@@ -80,7 +80,7 @@ pub fn breakdowns(conn: &Connection, threshold_days: u64) -> rusqlite::Result<Br
 /// the LanceDB directory tree. Missing files / dir = 0.
 pub fn index_bytes(data_dir: &Path) -> std::io::Result<IndexBytes> {
    fn file_size_or_zero(p: &Path) -> u64 {
-        std::fs::metadata(p).map(|m| m.len()).unwrap_or(0)
+        std::fs::metadata(p).map_or(0, |m| m.len())
    }
    fn dir_walk_sum(p: &Path) -> std::io::Result<u64> {
        if !p.exists() {
--- a/crates/kebab-store-sqlite/tests/contract_roundtrip.rs
+++ b/crates/kebab-store-sqlite/tests/contract_roundtrip.rs
@@ -57,6 +57,8 @@ fn document_and_chunks_round_trip_through_sqlite() {
        fs_ctime: asset.discovered_at,
        fs_mtime: asset.discovered_at,
        fallback_lang: Some("en".into()),
+        source_id: None,
+        fallback_trust_level: None,
    };
    let (mut metadata, _fm_span, _fm_warns) = parse_frontmatter(&bytes, &hints).unwrap();
    let (parsed_blocks, parse_warns) = parse_blocks(&bytes, 1).unwrap();
--- a/crates/kebab-store-sqlite/tests/idempotency.rs
+++ b/crates/kebab-store-sqlite/tests/idempotency.rs
@@ -45,6 +45,7 @@ fn make_metadata() -> Metadata {
        git_branch: None,
        git_commit: None,
        code_lang: None,
+        source_id: None,
    }
 }

--- a/crates/kebab-store-sqlite/tests/incremental_ingest.rs
+++ b/crates/kebab-store-sqlite/tests/incremental_ingest.rs
@@ -55,6 +55,7 @@ fn make_doc() -> CanonicalDocument {
        git_branch: None,
        git_commit: None,
        code_lang: None,
+        source_id: None,
    };
    CanonicalDocument {
        doc_id,
--- a/crates/kebab-store-sqlite/tests/list_docs.rs
+++ b/crates/kebab-store-sqlite/tests/list_docs.rs
@@ -58,6 +58,7 @@ fn make_doc(
        git_branch: None,
        git_commit: None,
        code_lang: None,
+        source_id: None,
    };
    let doc = CanonicalDocument {
        doc_id,
--- a/crates/kebab-tui/src/ask.rs
+++ b/crates/kebab-tui/src/ask.rs
@@ -598,8 +598,7 @@ fn spawn_ask_worker(state: &mut App) {
 fn make_conversation_id() -> String {
    let nanos = std::time::SystemTime::now()
        .duration_since(std::time::UNIX_EPOCH)
-        .map(|d| d.as_nanos())
-        .unwrap_or(0);
+        .map_or(0, |d| d.as_nanos());
    format!("conv_{nanos:032x}")
 }

--- a/crates/kebab-tui/src/ingest_progress.rs
+++ b/crates/kebab-tui/src/ingest_progress.rs
@@ -34,11 +34,10 @@ pub fn start_ingest(app: &mut App) -> anyhow::Result<()> {
        anyhow::bail!("ingest already running");
    }
    let cfg = app.config.clone();
-    let scope = SourceScope {
-        root: std::path::PathBuf::from(&cfg.workspace.root),
-        exclude: cfg.workspace.exclude.clone(),
-        ..Default::default()
-    };
+    // [[workspace.sources]]: leave `scope.root` empty so the app iterates
+    // every configured source (`config.resolved_sources()`), mirroring the
+    // CLI `kebab ingest` path. Each source carries its own merged exclude.
+    let scope = SourceScope::default();
    let (tx, rx) = mpsc::channel::<IngestEvent>();
    let cancel = Arc::new(AtomicBool::new(false));
    let cancel_for_worker = cancel.clone();
--- a/crates/kebab-tui/src/search.rs
+++ b/crates/kebab-tui/src/search.rs
@@ -304,10 +304,11 @@ pub fn handle_key_search(state: &mut App, key: KeyEvent) -> KeyOutcome {
            // `terminal.clear()` couldn't happen — leaving the
            // previous frame leaking through the new draw.
            let editor = std::env::var("EDITOR").unwrap_or_else(|_| "vi".into());
-            // `~/...` / `${XDG_…}` expansion via `kebab-config::expand_path`
-            // — same helper used by the markdown / image / PDF ingest
-            // paths (HOTFIXES 2026-05-02 P9-4 follow-up).
-            let workspace_root = kebab_config::expand_path(&state.config.workspace.root, "");
+            // [[workspace.sources]]: resolve the primary workspace root
+            // (first source / legacy `root`). `resolve_workspace_root` applies
+            // the same `~` / `${XDG_…}` / relative-to-config expansion as the
+            // markdown / image / PDF ingest paths (HOTFIXES 2026-05-02 P9-4).
+            let workspace_root = state.config.resolve_workspace_root();
            state.pending_editor = Some(crate::app::EditorRequest {
                citation: citation.unwrap(),
                editor_env: editor,
--- a/crates/kebab-tui/tests/ask.rs
+++ b/crates/kebab-tui/tests/ask.rs
@@ -19,7 +19,7 @@ use time::OffsetDateTime;
 fn fresh_app() -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-ask-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-ask-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-ask-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = Pane::Ask;
    // p9-fb-12 follow-up: mirror the run loop's auto-flip on pane
--- a/crates/kebab-tui/tests/cheatsheet.rs
+++ b/crates/kebab-tui/tests/cheatsheet.rs
@@ -12,7 +12,7 @@ use ratatui::layout::Rect;
 fn fresh_app(focus: Pane) -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-cheatsheet-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-cheatsheet-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-cheatsheet-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = focus;
    app
--- a/crates/kebab-tui/tests/inspect.rs
+++ b/crates/kebab-tui/tests/inspect.rs
@@ -23,7 +23,7 @@ use time::OffsetDateTime;
 fn fresh_app() -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-inspect-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-inspect-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-inspect-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = Pane::Inspect;
    app.inspect = Some(InspectState::default());
@@ -85,6 +85,7 @@ fn make_doc() -> CanonicalDocument {
            git_branch: None,
            git_commit: None,
            code_lang: None,
+            source_id: None,
        },
        provenance: Provenance {
            events: vec![ProvenanceEvent {
--- a/crates/kebab-tui/tests/mode.rs
+++ b/crates/kebab-tui/tests/mode.rs
@@ -9,7 +9,7 @@ use kebab_tui::{App, Mode, Pane, mode_intercept};
 fn fresh_app(focus: Pane) -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-mode-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-mode-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-mode-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = focus;
    app.mode = Mode::auto_for(focus);
--- a/crates/kebab-tui/tests/search.rs
+++ b/crates/kebab-tui/tests/search.rs
@@ -18,7 +18,7 @@ use std::path::Path;
 fn fresh_app() -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-search-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-search-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-search-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = Pane::Search;
    // p9-fb-12 follow-up: mirror the run loop's auto-flip — Search
--- a/crates/kebab-tui/tests/status_bar.rs
+++ b/crates/kebab-tui/tests/status_bar.rs
@@ -9,7 +9,7 @@ use ratatui::layout::Rect;
 fn fresh_app(focus: Pane) -> App {
    let mut config = Config::defaults();
    config.storage.data_dir = "/tmp/kebab-tui-status-bar-tests-noop".to_string();
-    config.workspace.root = "/tmp/kebab-tui-status-bar-tests-noop/workspace".to_string();
+    config.workspace.root = Some("/tmp/kebab-tui-status-bar-tests-noop/workspace".to_string());
    let mut app = App::new(config).expect("App::new");
    app.focus = focus;
    app
--- a/docs/ARCHITECTURE.md
+++ b/docs/ARCHITECTURE.md
@@ -34,6 +34,7 @@ Cargo workspace, 함수 호출 기반 모듈러 모놀리스. UI binary (`kebab-
 | RRF fusion_score | `[0, 1]` 정규화 — `2 / (k_rrf + 1)` 로 나눠 mode 간 비교 가능 (post-merge hotfix) |
 | ~~doc-side expansion 별칭 (v0.21.0)~~ | **제거됨 (v0.25.0, HOTFIXES 2026-06-03)** — 색인-시 청크당 LLM 별칭 생성 + 별칭 검색 채널을 완전히 제거. 별칭 ROI 음수(cross-lingual 은 e5-large 단독으로 충분, 기여는 설명형 +2 그룹뿐인데 대가가 청크당 색인-시 LLM). V013 마이그레이션이 `chunk_aliases_fts` + `chunks.aliases` DROP. 기존 KB 의 잔존 별칭 벡터는 검색 시 `strip_alias_suffix` 로 본문 chunk 에 매핑(graceful)되거나 `kebab reset` 으로 정리. spec: `docs/superpowers/specs/2026-06-03-remove-doc-expansion-spec.md`. |
 | 파생물 캐시 `derivation_cache` (V012, v0.21.0) | 비싼 ingest 파생물(embedding 벡터)을 청크 **내용 해시** 키로 SQLite 에 캐싱 → 재색인 시 내용 불변 청크는 재계산 skip. `cache_key = blake3(kind ‖ text_blake3 ‖ version_key)[:32]`; version_key 에 model/dimensions 포함 → §9 cascade 와 정합(버전 bump 시 자동 miss). 위치 기반 `chunk_id` 와 달리 내용이 같으면 문서·위치 무관 동일 키. 순수 가산 — `corpus_revision` bump 안 함, 손상/삭제돼도 정확성 영향 0(miss → 재계산). search/ask 는 `kebab.sqlite`+`lancedb` 만으로 동작하므로 외부 서버 색인 후 DB 만 복사하는 이식 워크플로 가능 (HOTFIXES 2026-05-31). (별칭 LLM 캐싱 kind 는 v0.25.0 에서 제거 — embedding kind 만 남음.) |
+| provenance 출처 필터 (v0.29.0) | 혼합 출처 KB 의 레버 = **질의 시 출처 필터링** (전역 trust 가중 아님). config `[[workspace.sources]]`(각 id/root/trust_level/source_type) → `documents.source_id` 컬럼(V014, additive·DEFAULT `'default'`) stamp + 검색 `--source <id>` / `--source-type <type>`(lexical+vector 두 site, OR). 단일 root 는 implicit `default` source 로 정규화(config v3→v4 `step_3_to_4` 미러). per-source trust/type 는 frontmatter 부재 시 기본값(우선순위 frontmatter > source 기본값 > Primary). **전역 trust 곱셈가중(weighted-RRF)은 반증** — A/B 에서 θ=0.85 만으로 incident MRR 0.918→0.340 절벽(점수 압축), 작은 오염 잡으려다 큰 개선 버리는 see-saw 라 빌드 안 함. 필터는 see-saw 없음. (HOTFIXES 2026-06-21) |
 | layout | XDG (`~/.local/share/kebab/`, `~/.config/kebab/`, …) |

 전체 frozen 설계는 [docs/superpowers/specs/2026-04-27-kebab-final-form-design.md](superpowers/specs/2026-04-27-kebab-final-form-design.md) 12 sections 참조.
@@ -219,7 +220,7 @@ kebab/
 │   ├── kebab-tui/                                     # Ratatui shell + Library 패널 (P9-1)
 │   ├── kebab-mcp/                                     # stdio MCP server — tools: schema, doctor, search, ask (P9-FB-30)
 │   └── kebab-cli/                                     # binary (P0 → 핫픽스로 --config flag wiring 강화)
-├── migrations/                                     # SQLite refinery V001..V012 (V012 = derivation_cache, v0.21.0)
+├── migrations/                                     # SQLite refinery V001..V014 (V012 = derivation_cache v0.21.0, V013 = drop chunk_aliases v0.25.0, V014 = documents.source_id v0.29.0)
 └── fixtures/                                       # 테스트 fixture 트리
 ```

--- a/docs/DOGFOOD.md
+++ b/docs/DOGFOOD.md
@@ -397,6 +397,22 @@ $KB search 'tokenizer' --mode lexical --json | jq '.hits | length' # ≥ 1 if co
 - `--lang` ISO code.
 - `--path-glob` workspace_path glob.

+### §2.4bis Source / provenance filters (`--source` / `--source-type`, v0.29.0)
+
+```bash
+# 출처 id 필터 ([[workspace.sources]] 의 id; 단일 root 는 "default").
+"$RELEASE_BIN" search --config "$DOGFOOD/config.toml" "query" --source jira --json | jq '.hits | length'
+# source_type 필터 (markdown/note/paper/reference/inbox).
+"$RELEASE_BIN" search --config "$DOGFOOD/config.toml" "query" --source-type reference,markdown --json
+```
+
+**verify**:
+- `--source` / `--source-type` repeatable + comma-sep, OR within.
+- lexical · vector · hybrid 모든 모드에 동일 적용 (직접 인덱스 컬럼 `documents.source_id` / `source_type`).
+- 모르는 값 → silently empty (no error).
+- 멀티소스 KB 측정: `--source wiki` 가 개념 질의 오염 회복(MRR 0.780→0.810), `--source jira` 가 incident 0.918→0.975 (HOTFIXES 2026-06-21).
+- trust precedence: `[[workspace.sources]]` 의 per-source `trust_level` 가 frontmatter 부재 시 적용 → `--trust-min primary` 와 조합 시 secondary source 배제.
+
 ### §2.5 Search pagination (p9-fb-34)

 ```bash
--- a/docs/SMOKE.md
+++ b/docs/SMOKE.md
@@ -290,6 +290,21 @@ kebab search "rust" --doc-id "<doc-id>" --tag rust --json
 Bad `--ingested-after` → `error.v1.code = config_invalid`, exit 2.
 Unknown `--media` value → silently empty (no error).

+### Source filters (`--source` / `--source-type`)
+
+````bash
+# 단일 root 워크스페이스는 implicit `default` source 로 정규화되므로
+# 모든 문서가 source_id="default" — 이 필터는 전체와 동일하다.
+kebab search "rust" --source default --json | jq '.hits | length'
+
+# source_type 필터 (frontmatter 의 source_type: 또는 source 기본값).
+kebab search "rust" --source-type markdown,reference --json | jq '.hits | length'
+````
+
+멀티소스 KB 는 `[[workspace.sources]]` 로 명명 source 를 선언하면
+`--source <id>` 로 출처를 좁힌다 (예: `--source jira` → jira 문서만).
+빈 값 = 무필터, 콤마/반복 = OR. 모르는 값 → silently empty (no error).
+
 ### Trace + stats (fb-37)

 Re-run a search with `--trace` to see per-stage candidate lists + timing:
--- a/docs/release-notes/v0.29.0-draft.md
+++ b/docs/release-notes/v0.29.0-draft.md
@@ -0,0 +1,104 @@
+---
+title: kebab v0.29.0 release notes (draft)
+created: 2026-06-21
+status: draft
+release_trigger:
+  - 신규 CLI flag `kebab search --source` / `--source-type` — pre-1.0 minor bump
+  - 신규 config `[[workspace.sources]]` + V014 migration (documents.source_id) — 인터페이스 추가
+  - config schema v3→v4 (단일 root → implicit default source 미러)
+---
+
+# kebab v0.29.0 — provenance 출처 필터: 멀티소스 + `--source`
+
+v0.28.0(config 스키마 재편) 후속 minor release. 혼합 출처 KB — 예컨대 위키
+문서와 jira 이슈를 한 KB 에 같이 넣은 경우 — 에서 **색인은 전부 하되 질의 시
+출처로 좁히는** provenance 레버를 추가한다. **기존 단일-폴더 사용자는 아무것도
+손대지 않아도 된다** — 단일 `workspace.root` 는 자동으로 `default` 라는 하나의
+source 로 정규화되고, 새 DB 컬럼은 기존 문서에 `default` 를 자동 채우는 additive
+마이그레이션이라 재색인이 발생하지 않는다.
+
+---
+
+## 변경 사실
+
+**1) 검색 출처 필터 2종.** `kebab search` 에 두 필터가 생겼다.
+
+| flag | 의미 |
+|---|---|
+| `--source <id>` | `[[workspace.sources]]` 에 선언한 source 의 id 로 필터 (예: `--source jira`) |
+| `--source-type <type>` | `markdown` / `note` / `paper` / `reference` / `inbox` 로 필터 |
+
+둘 다 반복(`--source a --source b`) 또는 콤마(`--source a,b`)로 여러 값을 줄 수
+있고 OR 로 묶인다. 빈 값 = 무필터. lexical · vector · hybrid **모든 검색 모드**에
+동일하게 적용된다(직접 인덱스 컬럼 — 추가 비용 거의 없음).
+
+**2) `[[workspace.sources]]` 멀티소스 config.** 단일 `[workspace] root` 대신
+여러 폴더를 명명 source 로 선언할 수 있다.
+
+```toml
+[[workspace.sources]]
+id = "notes"
+root = "~/KnowledgeBase/notes"
+
+[[workspace.sources]]
+id = "jira"
+root = "~/exports/jira"
+trust_level = "secondary"   # frontmatter 가 없을 때의 출처 기본 신뢰도
+source_type = "reference"   # frontmatter 가 없을 때의 출처 기본 타입
+```
+
+각 source 의 `id` 는 그 폴더에서 색인된 모든 문서에 stamp 되고
+(`documents.source_id` 컬럼), `--source <id>` 필터의 대상이 된다.
+`trust_level` / `source_type` 은 **출처 기본값**으로, 문서 frontmatter 가 해당
+필드를 지정하지 않을 때만 적용된다 (우선순위: **frontmatter > source 기본값 >
+하드코딩 Primary/Markdown**). `kebab ingest` 는 `--root` 를 주지 않으면 선언된
+모든 source 를 각자의 root + exclude 로 순회한다.
+
+## Trade-off — 왜 필터인가 (전역 trust 가중은 반증됨)
+
+"출처가 섞이면 신뢰도로 가중하면 되지 않나?" 를 통제 실험으로 검증했고,
+**전역 trust 곱셈가중은 반증됐다**. jira 를 docs KB 에 섞으면 개념 질의는 약하게
+오염(top-3 정답은 유지, rank1→2 강등)되지만 운영/이슈 질의는 크게 개선된다
+(jira_only hit@10 0/10 → 10/10). 그런데 jira 점수에 θ=0.85 만 곱해도 RAG 점수
+압축 때문에 incident MRR 이 0.918→0.340 으로 절벽 하락한다 — 작은 오염을 잡으려다
+큰 개선을 버리는 see-saw. 그래서 전역 가중은 **빌드하지 않았다**.
+
+올바른 레버는 **질의 시 출처 필터링**이다: 색인은 전부 해 두고(운영 질의는
+jira 가 답하게), 개념 질의에서만 `--source wiki` 로 좁힌다. see-saw 없이 양쪽을
+다 얻는다.
+
+## Mitigation — 기존 사용자 무영향 (재색인 0)
+
+- **단일 root 그대로 동작**: `[[workspace.sources]]` 를 선언하지 않으면 기존
+  `workspace.root` 가 implicit `default` source 로 정규화된다. 모든 문서가
+  `source_id = "default"`.
+- **V014 는 additive**: `documents.source_id` 컬럼은 `DEFAULT 'default'` 라 기존
+  row 가 자동으로 채워진다. 데이터 재작성·재색인·`corpus_revision` bump 없음.
+- **config v3→v4 자동 변환**: load 시 메모리에서 자동 변환(디스크 미변경),
+  `kebab config migrate` 로 파일 갱신 시 값·주석 보존 + 멱등. 단일 `root` 는
+  `[[workspace.sources]]` id=default 로 **미러**되며 기존 `root` 키도 그대로 남는다.
+
+도그푸딩(v0.29.0 release 빌드, 실험 corpus): 620 문서 / 0 error 색인,
+`source_id = {jira: 400, wiki: 220}`. trust precedence 실측 — jira 는 source
+기본값 secondary 라 `--trust-min primary` 시 0/6 노출, wiki 는 primary 유지.
+출처 필터 실측 — `--source wiki` 로 개념 질의 MRR 0.780→0.810(오염 회복),
+`--source jira` 로 incident 0.918→0.975.
+
+## Upgrade 절차
+
+1. 아무것도 안 해도 된다 — 기존 `config.toml` 과 KB 는 그대로 동작한다(단일 root
+   = `default` source, V014 자동 backfill).
+2. 출처를 나누고 싶으면 `kebab config migrate` 로 config 를 v4 로 갱신한 뒤
+   `[[workspace.sources]]` 블록을 손으로 추가하고 `kebab ingest --force-reingest`
+   로 각 문서에 새 `source_id` 를 stamp 한다. (단순 `ingest` 는 내용이 안 바뀐
+   문서를 skip 하므로, 기존 문서의 source_id 를 `default` 외 값으로 바꾸려면
+   `--force-reingest` 필요.)
+3. 검색에서 `--source <id>` / `--source-type <type>` 로 출처를 좁힌다.
+
+## Known limitations / 다음
+
+- **MCP search 도구**는 아직 `--source` / `--source-type` 를 노출하지 않는다
+  (CLI 전용). agent 용 MCP 필터 노출은 다음 additive 후보.
+- **`kebab list`** 출력(`doc_summary.v1`)에 `source_id` 가 아직 안 실린다.
+- **`kebab ask`** citation 에 provenance 라벨이 아직 없다 — 검색 필터는 되지만
+  답변 근거의 출처 표기는 다음 단계.
--- a/migrations/V014__documents_source_id.sql
+++ b/migrations/V014__documents_source_id.sql
@@ -0,0 +1,15 @@
+-- V014: [[workspace.sources]] multi-source support.
+--
+-- Adds `documents.source_id`: the id of the `[[workspace.sources]]` entry a
+-- document was ingested from. Single-root workspaces (and every pre-existing
+-- row) get the implicit `default` id via the column DEFAULT — so this is a
+-- backward-compatible additive migration (no data rewrite, no corpus_revision
+-- bump required for existing chunks/embeddings).
+--
+-- The DEFAULT 'default' literal is kept in sync with
+-- `kebab_config::DEFAULT_SOURCE_ID`. The index backs the `--source <id>`
+-- search filter (SearchFilters.source_id → `d.source_id IN (...)`).
+
+ALTER TABLE documents ADD COLUMN source_id TEXT NOT NULL DEFAULT 'default';
+
+CREATE INDEX idx_docs_source_id ON documents(source_id);
--- a/Show More
+++ b/Show More