context-mode 1.0.98 → 1.0.100

package/build/session/snapshot.js

@@ -312,6 +312,30 @@ function buildSkillsSection(skillEvents, searchTool) {
     ];
     return lines.join("\n");
 }
+function buildRolesSection(roleEvents, searchTool) {
+    if (roleEvents.length === 0)
+        return "";
+    const seen = new Set();
+    const summaryLines = [];
+    const queryTerms = [];
+    for (const ev of roleEvents) {
+        if (seen.has(ev.data))
+            continue;
+        seen.add(ev.data);
+        summaryLines.push(`    ${escapeXML(ev.data)}`);
+        queryTerms.push(ev.data);
+    }
+    if (summaryLines.length === 0)
+        return "";
+    const queries = buildQueries(queryTerms);
+    const lines = [
+        `  <roles count="${summaryLines.length}">`,
+        ...summaryLines,
+        toolCall(searchTool, queries),
+        `  </roles>`,
+    ];
+    return lines.join("\n");
+}
 function buildIntentSection(intentEvents) {
     if (intentEvents.length === 0)
         return "";
@@ -344,6 +368,7 @@ export function buildResumeSnapshot(events, opts) {
     const subagentEvents = [];
     const intentEvents = [];
     const skillEvents = [];
+    const roleEvents = [];
     for (const ev of events) {
         switch (ev.category) {
             case "file":
@@ -379,6 +404,9 @@ export function buildResumeSnapshot(events, opts) {
             case "skill":
                 skillEvents.push(ev);
                 break;
+            case "role":
+                roleEvents.push(ev);
+                break;
         }
     }
     // ── Build all sections ──
@@ -417,6 +445,9 @@ export function buildResumeSnapshot(events, opts) {
     const skills = buildSkillsSection(skillEvents, searchTool);
     if (skills)
         sections.push(skills);
+    const roles = buildRolesSection(roleEvents, searchTool);
+    if (roles)
+        sections.push(roles);
     const intent = buildIntentSection(intentEvents);
     if (intent)
         sections.push(intent);

package/build/store.js

@@ -218,6 +218,40 @@ function findAllPositions(text, term) {
     }
     return positions;
 }
+/**
+ * Count matched adjacent pairs across consecutive query terms.
+ * For each pair (term[i], term[i+1]), pairs each left position with at most one
+ * right position whose offset falls within `gap` chars of `p + len(term[i])`.
+ * `positionLists` must be sorted ascending (output of `findAllPositions` is).
+ * Each right position is consumed by at most one left, so `"foo foo bar"`
+ * counts 1 pair, not 2 — matches IR phrase-occurrence intent and avoids
+ * inflating boosts for repeated-token queries.
+ * Used by reranker to layer a frequency signal on top of minSpan proximity:
+ * 30-char gap covers natural prose without rewarding distant matches.
+ */
+function countAdjacentPairs(positionLists, terms, gap = 30) {
+    if (positionLists.length < 2 || terms.length < 2)
+        return 0;
+    let total = 0;
+    const pairs = Math.min(positionLists.length, terms.length) - 1;
+    for (let i = 0; i < pairs; i++) {
+        const left = positionLists[i];
+        const right = positionLists[i + 1];
+        const leftLen = terms[i].length;
+        let j = 0;
+        for (const p of left) {
+            const minStart = p + leftLen;
+            const maxStart = minStart + gap;
+            while (j < right.length && right[j] < minStart)
+                j++;
+            if (j < right.length && right[j] <= maxStart) {
+                total++;
+                j++;
+            }
+        }
+    }
+    return total;
+}
 /**
  * Find minimum span (window) covering at least one position from each list.
  * Uses a sweep-line approach: advance the pointer at the current minimum.
@@ -368,6 +402,10 @@ export class ContentStore {
         content,
         source_id UNINDEXED,
         content_type UNINDEXED,
+        source_category UNINDEXED,
+        session_id UNINDEXED,
+        event_id UNINDEXED,
+        timestamp UNINDEXED,
         tokenize='porter unicode61'
       );
 
@@ -376,6 +414,10 @@ export class ContentStore {
         content,
         source_id UNINDEXED,
         content_type UNINDEXED,
+        source_category UNINDEXED,
+        session_id UNINDEXED,
+        event_id UNINDEXED,
+        timestamp UNINDEXED,
         tokenize='trigram'
       );
 
@@ -385,6 +427,47 @@ export class ContentStore {
 
       CREATE INDEX IF NOT EXISTS idx_sources_label ON sources(label);
     `);
+        // FTS5 schema migration: old schema (4 cols) → new schema (8 cols).
+        // FTS5 virtual tables do not support ALTER TABLE ADD COLUMN, so we must
+        // DROP + re-CREATE. Detection: check for sentinel column `source_category`
+        // via pragma_table_xinfo. Three states:
+        //   1. No table          → CREATE above handled it (fresh DB)
+        //   2. Old schema (4 cols) → DROP + CREATE new
+        //   3. New schema (8 cols) → do nothing
+        try {
+            const cols = this.#db.prepare("SELECT name FROM pragma_table_xinfo('chunks')").all();
+            const colNames = new Set(cols.map(c => c.name));
+            if (cols.length > 0 && !colNames.has("source_category")) {
+                // Old schema detected — drop both FTS5 tables and re-create with new columns
+                this.#db.exec("DROP TABLE IF EXISTS chunks");
+                this.#db.exec("DROP TABLE IF EXISTS chunks_trigram");
+                this.#db.exec(`
+          CREATE VIRTUAL TABLE chunks USING fts5(
+            title,
+            content,
+            source_id UNINDEXED,
+            content_type UNINDEXED,
+            source_category UNINDEXED,
+            session_id UNINDEXED,
+            event_id UNINDEXED,
+            timestamp UNINDEXED,
+            tokenize='porter unicode61'
+          );
+          CREATE VIRTUAL TABLE chunks_trigram USING fts5(
+            title,
+            content,
+            source_id UNINDEXED,
+            content_type UNINDEXED,
+            source_category UNINDEXED,
+            session_id UNINDEXED,
+            event_id UNINDEXED,
+            timestamp UNINDEXED,
+            tokenize='trigram'
+          );
+        `);
+            }
+        }
+        catch { /* pragma_table_xinfo may fail if table doesn't exist yet — safe to ignore */ }
         // Stale detection columns — safe for existing DBs (ALTER is O(1) in SQLite)
         try {
             this.#db.exec("ALTER TABLE sources ADD COLUMN file_path TEXT");
@@ -399,8 +482,8 @@ export class ContentStore {
         // Write path
         this.#stmtInsertSourceEmpty = this.#db.prepare("INSERT INTO sources (label, chunk_count, code_chunk_count, file_path, content_hash) VALUES (?, 0, 0, ?, ?)");
         this.#stmtInsertSource = this.#db.prepare("INSERT INTO sources (label, chunk_count, code_chunk_count, file_path, content_hash) VALUES (?, ?, ?, ?, ?)");
-        this.#stmtInsertChunk = this.#db.prepare("INSERT INTO chunks (title, content, source_id, content_type) VALUES (?, ?, ?, ?)");
-        this.#stmtInsertChunkTrigram = this.#db.prepare("INSERT INTO chunks_trigram (title, content, source_id, content_type) VALUES (?, ?, ?, ?)");
+        this.#stmtInsertChunk = this.#db.prepare("INSERT INTO chunks (title, content, source_id, content_type, source_category, session_id, event_id, timestamp) VALUES (?, ?, ?, ?, ?, ?, ?, ?)");
+        this.#stmtInsertChunkTrigram = this.#db.prepare("INSERT INTO chunks_trigram (title, content, source_id, content_type, source_category, session_id, event_id, timestamp) VALUES (?, ?, ?, ?, ?, ?, ?, ?)");
         this.#stmtInsertVocab = this.#db.prepare("INSERT OR IGNORE INTO vocabulary (word) VALUES (?)");
         // Dedup path: delete previous source with same label before re-indexing
         // Prevents stale outputs from accumulating in iterative workflows (build-fix-build)
@@ -413,6 +496,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -427,6 +511,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -441,6 +526,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -455,6 +541,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -469,6 +556,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -483,6 +571,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -498,6 +587,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -512,6 +602,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -526,6 +617,7 @@ export class ContentStore {
         chunks.title,
         chunks.content,
         chunks.content_type,
+        chunks.timestamp,
         sources.label,
         bm25(chunks, 5.0, 1.0) AS rank,
         highlight(chunks, 1, char(2), char(3)) AS highlighted
@@ -540,6 +632,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -554,6 +647,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -568,6 +662,7 @@ export class ContentStore {
         chunks_trigram.title,
         chunks_trigram.content,
         chunks_trigram.content_type,
+        chunks_trigram.timestamp,
         sources.label,
         bm25(chunks_trigram, 5.0, 1.0) AS rank,
         highlight(chunks_trigram, 1, char(2), char(3)) AS highlighted
@@ -603,10 +698,16 @@ export class ContentStore {
     // ── Index ──
     index(options) {
         const { content, path, source } = options;
-        if (!content && !path) {
+        // Treat empty string as "no content" so an empty `content` paired with a
+        // valid `path` falls back to reading the file. Some MCP clients
+        // materialize optional string fields as `""` and the previous
+        // `content ?? readFileSync(path)` kept the empty string, indexing 0
+        // chunks. See issue #350.
+        const hasContent = typeof content === "string" && content.length > 0;
+        if (!hasContent && !path) {
             throw new Error("Either content or path must be provided");
         }
-        const text = content ?? readFileSync(path, "utf-8");
+        const text = hasContent ? content : readFileSync(path, "utf-8");
         const label = source ?? path ?? "untitled";
         const chunks = this.#chunkMarkdown(text);
         // Stale detection: store file_path + SHA-256 for file-backed sources
@@ -674,10 +775,11 @@ export class ContentStore {
             }
             const info = this.#stmtInsertSource.run(label, chunks.length, codeChunks, filePath ?? null, contentHash ?? null);
             const sourceId = Number(info.lastInsertRowid);
+            const now = new Date().toISOString();
             for (const chunk of chunks) {
                 const ct = chunk.hasCode ? "code" : "prose";
-                this.#stmtInsertChunk.run(chunk.title, chunk.content, sourceId, ct);
-                this.#stmtInsertChunkTrigram.run(chunk.title, chunk.content, sourceId, ct);
+                this.#stmtInsertChunk.run(chunk.title, chunk.content, sourceId, ct, null, null, null, now);
+                this.#stmtInsertChunkTrigram.run(chunk.title, chunk.content, sourceId, ct, null, null, null, now);
             }
             return sourceId;
         });
@@ -708,6 +810,7 @@ export class ContentStore {
             rank: r.rank,
             contentType: r.content_type,
             highlighted: r.highlighted,
+            timestamp: r.timestamp ?? undefined,
         }));
     }
     #sourceFilterParam(source, sourceMatchMode) {
@@ -859,17 +962,24 @@ export class ContentStore {
             const titleHits = terms.filter((t) => titleLower.includes(t)).length;
             const titleWeight = r.contentType === "code" ? 0.6 : 0.3;
             const titleBoost = titleHits > 0 ? titleWeight * (titleHits / terms.length) : 0;
-            // Proximity boost for multi-term queries
+            // Proximity boost for multi-term queries. minSpan picks the single
+            // tightest window — frequency doesn't move it, so a long doc with one
+            // tight occurrence outranks a short doc with several. Phrase-frequency
+            // reward layers a saturating frequency signal on top: cap 0.5 (below
+            // proximity max ≈1.0, in title-boost range), saturates at 4 hits.
             let proximityBoost = 0;
+            let phraseBoost = 0;
             if (terms.length >= 2) {
                 const content = r.content.toLowerCase();
                 const positions = terms.map((t) => findAllPositions(content, t));
                 if (!positions.some((p) => p.length === 0)) {
                     const minSpan = findMinSpan(positions);
                     proximityBoost = 1 / (1 + minSpan / Math.max(content.length, 1));
+                    const adjacentPairs = countAdjacentPairs(positions, terms);
+                    phraseBoost = 0.5 * Math.min(1, adjacentPairs / 4);
                 }
             }
-            return { result: r, boost: titleBoost + proximityBoost };
+            return { result: r, boost: titleBoost + proximityBoost + phraseBoost };
         })
             .sort((a, b) => b.boost - a.boost || a.result.rank - b.result.rank)
             .map(({ result }) => result);

package/build/types.d.ts

@@ -74,6 +74,7 @@ export interface SearchResult {
     contentType: "code" | "prose";
     matchLayer?: "porter" | "trigram" | "fuzzy" | "rrf" | "rrf-fuzzy";
     highlighted?: string;
+    timestamp?: string;
 }
 /**
  * Aggregate statistics for a ContentStore instance.