claude-mem-lite 2.88.0 → 2.90.0

package/.claude-plugin/marketplace.json

@@ -10,9 +10,9 @@
   "plugins": [
     {
       "name": "claude-mem-lite",
-      "version": "2.88.0",
+      "version": "2.90.0",
       "source": "./",
-      "description": "Persistent long-term memory for Claude Code via MCP — captures coding decisions, bugfixes, and context across sessions. Hybrid FTS5 + TF-IDF search with episode batching. Single SQLite DB, no external services. Alternative to claude-mem with 600x lower cost."
+      "description": "Persistent long-term memory for Claude Code via MCP — captures coding decisions, bugfixes, and context across sessions. Hybrid FTS5 + TF-IDF search with episode batching. Single SQLite DB, no external services. A lighter, lower-cost alternative to claude-mem (episode batching + a smaller model; cost savings are an internal estimate, not a measured benchmark)."
     }
   ]
 }

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "claude-mem-lite",
-  "version": "2.88.0",
-  "description": "Persistent long-term memory for Claude Code via MCP — captures coding decisions, bugfixes, and context across sessions. Hybrid FTS5 + TF-IDF search with episode batching. Single SQLite DB, no external services. Alternative to claude-mem with 600x lower cost.",
+  "version": "2.90.0",
+  "description": "Persistent long-term memory for Claude Code via MCP — captures coding decisions, bugfixes, and context across sessions. Hybrid FTS5 + TF-IDF search with episode batching. Single SQLite DB, no external services. A lighter, lower-cost alternative to claude-mem (episode batching + a smaller model; cost savings are an internal estimate, not a measured benchmark).",
   "author": {
     "name": "sdsrss"
   },

package/README.md

@@ -4,7 +4,7 @@
 
 `claude-mem-lite` is a **persistent memory** (also called *long-term memory* or *cross-session context*) system for **[Claude Code](https://docs.anthropic.com/en/docs/claude-code)** — Anthropic's CLI coding agent. It runs as an **[MCP](https://modelcontextprotocol.io/) server** plus a set of Claude Code hooks, automatically capturing coding observations, decisions, and bug fixes during sessions, then providing hybrid full-text + semantic search to recall them later.
 
-Compared to general-purpose LLM memory frameworks like [`mem0`](https://github.com/mem0ai/mem0) or the MCP reference [`memory`](https://github.com/modelcontextprotocol/servers/tree/main/src/memory) server, claude-mem-lite is purpose-built for Claude Code's hook lifecycle: episode batching cuts LLM calls 7–10× vs the original [claude-mem](https://github.com/thedotmack/claude-mem) (600× lower total cost), and the hybrid FTS5 + TF-IDF retriever benchmarks at 0.88 Recall@10 / 0.96 Precision@10.
+Compared to general-purpose LLM memory frameworks like [`mem0`](https://github.com/mem0ai/mem0) or the MCP reference [`memory`](https://github.com/modelcontextprotocol/servers/tree/main/src/memory) server, claude-mem-lite is purpose-built for Claude Code's hook lifecycle: episode batching cuts LLM calls 7–10× vs the original [claude-mem](https://github.com/thedotmack/claude-mem) (an estimated ~600× lower total cost — see the cost model below; this is an architecture estimate, not a measured benchmark), while the hybrid FTS5 + TF-IDF retriever benchmarks at 0.88 Recall@10 / 0.96 Precision@10.
 
 > 中文简介：claude-mem-lite 是 Claude Code 的轻量级**持久化记忆 / 长期记忆 / 跨会话上下文**插件，基于 MCP 协议 + 钩子机制，自动捕获编码会话中的决策、修复和上下文，并通过 FTS5 + TF-IDF 混合检索召回。详见 [中文 README](README.zh-CN.md)。
 
@@ -29,15 +29,17 @@ A ground-up redesign of [claude-mem](https://github.com/thedotmack/claude-mem),
 
 ### Token & cost efficiency
 
-For a typical 50-tool-call session:
+For a typical 50-tool-call session (illustrative cost model — the ratios below are
+architecture estimates derived from batch size, token counts, and model pricing, **not**
+a measured end-to-end benchmark):
 
-| | claude-mem | claude-mem-lite | Ratio |
+| | claude-mem | claude-mem-lite | Ratio (estimated) |
 |---|---|---|---|
-| LLM calls | ~50 (every tool use) | ~5-8 (per episode) | **7-10x fewer** |
-| Tokens per call | 1,000-5,000 (raw JSON + history) | 200-500 (summaries only) | **5-10x smaller** |
-| Total tokens | ~100K-250K | ~1K-4K | **50-100x less** |
-| Model cost | Sonnet ($3/$15 per M) | Haiku ($0.25/$1.25 per M) | **12x cheaper** |
-| Combined savings | | | **600x+ lower cost** |
+| LLM calls | ~50 (every tool use) | ~5-8 (per episode) | **~7-10x fewer** |
+| Tokens per call | 1,000-5,000 (raw JSON + history) | 200-500 (summaries only) | **~5-10x smaller** |
+| Total tokens | ~100K-250K | ~1K-4K | **~50-100x less** |
+| Model cost | Sonnet ($3/$15 per M) | Haiku ($0.25/$1.25 per M) | **~12x cheaper** |
+| Combined savings | | | **~600x lower cost (estimated)** |
 
 ### Quality comparison
 
@@ -681,7 +683,7 @@ No. Claude Code's `CLAUDE.md` and `MEMORY.md` files act as static instruction me
 
 ### Why "lite"? What did the original claude-mem do differently?
 
-The original called an LLM on every tool use with raw JSON inputs. claude-mem-lite batches 5–10 operations per LLM call, uses a smaller model (Haiku), and runs a deterministic code-level filter before sending anything to the model. Net result: ~600× lower cost with equivalent search quality. See the [Architecture comparison](#architecture-comparison) above.
+The original called an LLM on every tool use with raw JSON inputs. claude-mem-lite batches 5–10 operations per LLM call, uses a smaller model (Haiku), and runs a deterministic code-level filter before sending anything to the model. Net result: an estimated ~600× lower cost (an architecture estimate from the cost model above, not a measured benchmark) with equivalent search quality. See the [Architecture comparison](#architecture-comparison) above.
 
 ### Does this work cross-project? Cross-machine?
 

package/README.zh-CN.md

@@ -4,7 +4,7 @@
 
 `claude-mem-lite` 是 **[Claude Code](https://docs.anthropic.com/en/docs/claude-code)**（Anthropic 官方 CLI 编程代理）的 **持久化记忆系统**（也称 **长期记忆 / 跨会话上下文 / Claude Code 记忆插件**）。它以 **[MCP](https://modelcontextprotocol.io/) 服务器** + Claude Code 钩子（hooks）的形式运行，在编码会话中自动捕获观察记录、决策、bug 修复，并通过 FTS5 全文检索 + TF-IDF 向量的混合检索召回历史上下文。
 
-与 [`mem0`](https://github.com/mem0ai/mem0)、MCP 官方参考实现的 [`memory`](https://github.com/modelcontextprotocol/servers/tree/main/src/memory) 服务器等通用 LLM 记忆框架相比，claude-mem-lite 专为 Claude Code 的钩子生命周期定制：episode 批处理把 LLM 调用量相比原版 [claude-mem](https://github.com/thedotmack/claude-mem) 减少 7-10 倍（综合成本下降 600 倍），FTS5 + TF-IDF 混合检索在 30 个查询的基准上达到 **Recall@10 = 0.88 / Precision@10 = 0.96**。
+与 [`mem0`](https://github.com/mem0ai/mem0)、MCP 官方参考实现的 [`memory`](https://github.com/modelcontextprotocol/servers/tree/main/src/memory) 服务器等通用 LLM 记忆框架相比，claude-mem-lite 专为 Claude Code 的钩子生命周期定制：episode 批处理把 LLM 调用量相比原版 [claude-mem](https://github.com/thedotmack/claude-mem) 减少 7-10 倍（综合成本估算下降约 600 倍 —— 见下方成本模型，属架构估算而非实测基准）；FTS5 + TF-IDF 混合检索在 30 个查询的基准上达到 **Recall@10 = 0.88 / Precision@10 = 0.96**。
 
 无需外部服务。单一 SQLite 数据库。开销极低。
 
@@ -27,15 +27,15 @@
 
 ### Token 与成本效率
 
-以典型的 50 次工具调用的会话为例：
+以典型的 50 次工具调用的会话为例（成本模型示意 —— 下列比率由批大小、token 量与模型定价**估算**得出，并非端到端实测）：
 
-| | claude-mem | claude-mem-lite | 比率 |
+| | claude-mem | claude-mem-lite | 比率（估算） |
 |---|---|---|---|
-| LLM 调用次数 | ~50（每次工具使用） | ~5-8（按 episode） | **减少 7-10 倍** |
-| 每次调用 token | 1,000-5,000（原始 JSON + 历史） | 200-500（仅摘要） | **减少 5-10 倍** |
-| 总 token 量 | ~100K-250K | ~1K-4K | **减少 50-100 倍** |
-| 模型成本 | Sonnet ($3/$15 每百万) | Haiku ($0.25/$1.25 每百万) | **便宜 12 倍** |
-| 综合节省 | | | **成本降低 600 倍+** |
+| LLM 调用次数 | ~50（每次工具使用） | ~5-8（按 episode） | **约减少 7-10 倍** |
+| 每次调用 token | 1,000-5,000（原始 JSON + 历史） | 200-500（仅摘要） | **约减少 5-10 倍** |
+| 总 token 量 | ~100K-250K | ~1K-4K | **约减少 50-100 倍** |
+| 模型成本 | Sonnet ($3/$15 每百万) | Haiku ($0.25/$1.25 每百万) | **约便宜 12 倍** |
+| 综合节省 | | | **成本降低约 600 倍（估算）** |
 
 ### 质量对比
 

package/cli/activity.mjs

@@ -11,6 +11,7 @@
 import { inferProject } from '../utils.mjs';
 import { resolveProject } from '../project-utils.mjs';
 import { parseArgs, out, fail } from './common.mjs';
+import { parseIntFlag } from '../lib/cli-flags.mjs';
 
 function formatActivityResults(rows) {
   if (!rows || rows.length === 0) return '(no events)';
@@ -77,17 +78,20 @@ export async function cmdActivity(db, args) {
       fail(`[mem] activity search: invalid --type "${type}". Valid: ${[...VALID_EVENT_TYPES].join(', ')}`);
       return;
     }
-    const limit = flags.limit !== undefined ? parseInt(flags.limit, 10) : 10;
+    const limit = parseIntFlag(flags.limit, { name: '--limit', defaultValue: 10, max: 1000 });
     const rows = searchEvents(db, q, { project, type, limit });
     out(formatActivityResults(rows));
     return;
   }
 
   if (sub === 'recent') {
-    // Accept either `activity recent 5` or `activity recent --limit 5`.
-    const posLimit = positional.length > 0 ? parseInt(positional[0], 10) : NaN;
-    const flagLimit = flags.limit !== undefined ? parseInt(flags.limit, 10) : NaN;
-    const limit = Number.isFinite(posLimit) ? posLimit : (Number.isFinite(flagLimit) ? flagLimit : 20);
+    // Accept either `activity recent 5` or `activity recent --limit 5`. Both routed
+    // through parseIntFlag so garbage ("2abc"), negatives (SQLite LIMIT -1 = UNLIMITED
+    // full-table dump), and uncapped huge values warn + clamp to default/max, matching
+    // the search/recent/browse siblings.
+    const limit = positional.length > 0
+      ? parseIntFlag(positional[0], { name: 'count', defaultValue: 20, max: 1000 })
+      : parseIntFlag(flags.limit, { name: '--limit', defaultValue: 20, max: 1000 });
     const type = flags.type || null;
     if (type !== null && !VALID_EVENT_TYPES.has(type)) {
       fail(`[mem] activity recent: invalid --type "${type}". Valid: ${[...VALID_EVENT_TYPES].join(', ')}`);

package/cli.mjs

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-const CLI_COMMANDS = new Set(['search', 'recent', 'recall', 'get', 'timeline', 'save', 'stats', 'context', 'browse', 'citation-stats', 'delete', 'update', 'export', 'compress', 'maintain', 'optimize', 'fts-check', 'registry', 'import', 'import-jsonl', 'enrich', 'activity', 'adopt', 'unadopt', 'memdir-audit', 'defer', 'help']);
+const CLI_COMMANDS = new Set(['search', 'recent', 'recall', 'get', 'timeline', 'save', 'stats', 'context', 'browse', 'citation-stats', 'delete', 'update', 'export', 'restore', 'compress', 'maintain', 'optimize', 'fts-check', 'registry', 'import', 'import-jsonl', 'enrich', 'activity', 'adopt', 'unadopt', 'memdir-audit', 'defer', 'help']);
 const INSTALL_COMMANDS = new Set(['install', 'uninstall', 'status', 'doctor', 'cleanup', 'cleanup-hooks', 'self-update', 'repair', 'release']);
 
 const cmd = process.argv[2];
@@ -13,14 +13,16 @@ if (cmd === '--version' || cmd === '-v') {
 } else if (cmd === '--help' || cmd === '-h') {
   const { run } = await import('./mem-cli.mjs');
   await run(['help']);
-} else if (cmd === 'doctor' && process.argv.slice(3).some(a => a.startsWith('--') && a.length > 2)) {
-  // Per #8217 single-source-of-truth: any flagged `doctor --X` is a DB-layer
-  // inspection tool (--benchmark, --metrics, --session-audit, future flags)
-  // and routes to mem-cli. Plain `doctor` (no flags) keeps running the
-  // install health-check below — adding a new flag in cli/doctor.mjs no
-  // longer requires touching this enumeration. The `length > 2` guard
-  // ignores a bare `--` (POSIX end-of-options separator) so `doctor --`
-  // continues to route to install.mjs, not mem-cli.
+} else if (cmd === 'doctor' && process.argv.slice(3).some(a => a === '--benchmark' || a === '--metrics' || a === '--session-audit')) {
+  // Per #8217: the DB-layer doctor modes (--benchmark / --metrics / --session-audit,
+  // each implemented in cli/doctor.mjs) route to mem-cli. Everything else — plain
+  // `doctor`, `doctor --` (POSIX end-of-options), and `doctor --json` — stays with
+  // install.mjs's health-check, which OWNS --json (install.mjs doctor() line ~1216).
+  // Pre-fix the router forwarded ANY flagged `doctor --X` to mem-cli, so the documented
+  // `doctor --json` (install health JSON, advertised in install.mjs usage) was shadowed
+  // and rejected by cli/doctor.mjs. Gating on the three DB-layer flags keeps --json
+  // (and any future install-doctor flag) on the install path. Adding a NEW DB-layer
+  // mode requires extending this list — a deliberate trade for a working --json.
   const { run } = await import('./mem-cli.mjs');
   await run(process.argv.slice(2));
 } else if (CLI_COMMANDS.has(cmd)) {

package/haiku-client.mjs

@@ -20,6 +20,14 @@ const MODEL_MAP = {
   sonnet: 'claude-sonnet-4-5-20250929',
 };
 
+// Every background LLM call here is fixed-schema extraction / classification
+// (episode→JSON, type/merge classification, synonym + metadata extraction) whose
+// output is consumed deterministically (JSON.parse, MinHash dedup). Pin temperature
+// to 0 so the provider default (~1.0) doesn't inject wording variance that breaks
+// JSON parsing or defeats the wording-sensitive MinHash near-duplicate detector.
+// A call that genuinely needs sampling can pass opts.temperature to override.
+const DEFAULT_LLM_TEMPERATURE = 0;
+
 /**
  * Resolve the LLM model to use for background calls.
  * Reads CLAUDE_MEM_MODEL env var, defaults to 'haiku'.
@@ -143,7 +151,7 @@ export function flattenForCLI(input) {
  * @param {number} [opts.maxTokens=500] Max tokens in response
  * @returns {Promise<{text: string}|null>} Response or null on failure
  */
-export async function callHaiku(prompt, { timeout = 10000, maxTokens = 500 } = {}) {
+export async function callHaiku(prompt, { timeout = 10000, maxTokens = 500, temperature = DEFAULT_LLM_TEMPERATURE } = {}) {
   if (!prompt) return null;
 
   const mode = detectMode();
@@ -160,8 +168,8 @@ export async function callHaiku(prompt, { timeout = 10000, maxTokens = 500 } = {
   let primary = null;
   try {
     primary = mode === 'api'
-      ? await callHaikuAPI(prompt, { timeout, maxTokens })
-      : await callOpenRouterAPI(prompt, resolveModel().cli, { timeout, maxTokens });
+      ? await callHaikuAPI(prompt, { timeout, maxTokens, temperature })
+      : await callOpenRouterAPI(prompt, resolveModel().cli, { timeout, maxTokens, temperature });
   } catch (e) {
     debugCatch(e, `callHaiku:${mode}`);
   }
@@ -198,7 +206,7 @@ export async function callHaikuJSON(prompt, opts) {
  * @param {number} [opts.maxTokens=1000] Max tokens in response
  * @returns {Promise<{text: string}|null>} Response or null on failure
  */
-export async function callLLMWithModel(prompt, model = 'haiku', { timeout = 15000, maxTokens = 1000 } = {}) {
+export async function callLLMWithModel(prompt, model = 'haiku', { timeout = 15000, maxTokens = 1000, temperature = DEFAULT_LLM_TEMPERATURE } = {}) {
   if (!prompt) return null;
   const resolvedModel = MODEL_MAP[model] ? model : 'haiku';
   const mode = detectMode();
@@ -214,8 +222,8 @@ export async function callLLMWithModel(prompt, model = 'haiku', { timeout = 1500
   let primary = null;
   try {
     primary = mode === 'api'
-      ? await callModelAPI(prompt, resolvedModel, { timeout, maxTokens })
-      : await callOpenRouterAPI(prompt, resolvedModel, { timeout, maxTokens });
+      ? await callModelAPI(prompt, resolvedModel, { timeout, maxTokens, temperature })
+      : await callOpenRouterAPI(prompt, resolvedModel, { timeout, maxTokens, temperature });
   } catch (e) {
     debugCatch(e, `callLLMWithModel:${mode}:${resolvedModel}`);
   }
@@ -239,7 +247,7 @@ export async function callModelJSON(prompt, model = 'haiku', opts) {
   return parseJsonFromLLM(result.text);
 }
 
-async function callModelAPI(prompt, model, { timeout, maxTokens }) {
+async function callModelAPI(prompt, model, { timeout, maxTokens, temperature = DEFAULT_LLM_TEMPERATURE }) {
   const apiKey = process.env.ANTHROPIC_API_KEY;
   if (!apiKey) return null;
 
@@ -252,6 +260,7 @@ async function callModelAPI(prompt, model, { timeout, maxTokens }) {
     const body = {
       model: modelId,
       max_tokens: maxTokens,
+      temperature,
       messages: [{ role: 'user', content: user }],
     };
     // System slot is constant per call type (instructions, schema, type taxonomy)
@@ -312,7 +321,7 @@ function callModelCLI(prompt, model, { timeout }) {
 
 // ─── API Mode ────────────────────────────────────────────────────────────────
 
-async function callHaikuAPI(prompt, { timeout, maxTokens }) {
+async function callHaikuAPI(prompt, { timeout, maxTokens, temperature = DEFAULT_LLM_TEMPERATURE }) {
   const apiKey = process.env.ANTHROPIC_API_KEY;
   if (!apiKey) return null;
 
@@ -325,6 +334,7 @@ async function callHaikuAPI(prompt, { timeout, maxTokens }) {
     const body = {
       model: modelId,
       max_tokens: maxTokens,
+      temperature,
       messages: [{ role: 'user', content: user }],
     };
     // See callModelAPI: cache_control on the constant system slot.
@@ -365,7 +375,7 @@ async function callHaikuAPI(prompt, { timeout, maxTokens }) {
 // `cache_control` field has no OpenAI-format equivalent and is omitted.
 // `tier` is the resolved model tier ('haiku'|'sonnet'); OPENROUTER_MODEL can
 // override the resulting slug entirely (see resolveOpenRouterModel).
-async function callOpenRouterAPI(prompt, tier, { timeout, maxTokens }) {
+async function callOpenRouterAPI(prompt, tier, { timeout, maxTokens, temperature = DEFAULT_LLM_TEMPERATURE }) {
   const apiKey = process.env.OPENROUTER_API_KEY;
   if (!apiKey) return null;
 
@@ -387,7 +397,7 @@ async function callOpenRouterAPI(prompt, tier, { timeout, maxTokens }) {
         // Optional OpenRouter attribution headers (ignored by the API if absent).
         'X-Title': 'claude-mem-lite',
       },
-      body: JSON.stringify({ model, max_tokens: maxTokens, messages }),
+      body: JSON.stringify({ model, max_tokens: maxTokens, temperature, messages }),
       signal: controller.signal,
     });
 

package/hook-handoff.mjs

@@ -32,12 +32,26 @@ import * as taskReaderModule from './lib/task-reader.mjs';
  * @param {string|null} [scopeSessionId=null] CC UUID for session_handoffs.session_id column
  */
 export function buildAndSaveHandoff(db, sessionId, project, type, episodeSnapshot, scopeSessionId = null) {
-  // 1. Working objective — from user prompts
-  const prompts = db.prepare(`
-    SELECT prompt_text FROM user_prompts
-    WHERE content_session_id = ?
-    ORDER BY prompt_number ASC LIMIT 5
-  `).all(sessionId);
+  // 1. Working objective — from user prompts.
+  // D#26: getSessionId() is project-scoped, so multiple CC sessions in one project
+  // share `content_session_id`. When a genuine CC scope is passed (scopeSessionId is
+  // the CC UUID, i.e. differs from the mem-internal sessionId), filter to THIS CC
+  // session's prompts so working_on doesn't merge concurrent/sequential sessions.
+  // `OR cc_session_id IS NULL` keeps legacy rows + non-CC/no-stdin invocations. When
+  // scopeSessionId is absent or == sessionId (legacy/test/no-stdin), fall back to the
+  // unfiltered query (identical to pre-D#26 behavior).
+  const ccScope = scopeSessionId && scopeSessionId !== sessionId ? scopeSessionId : null;
+  const prompts = ccScope
+    ? db.prepare(`
+        SELECT prompt_text FROM user_prompts
+        WHERE content_session_id = ? AND (cc_session_id = ? OR cc_session_id IS NULL)
+        ORDER BY prompt_number ASC LIMIT 5
+      `).all(sessionId, ccScope)
+    : db.prepare(`
+        SELECT prompt_text FROM user_prompts
+        WHERE content_session_id = ?
+        ORDER BY prompt_number ASC LIMIT 5
+      `).all(sessionId);
   if (prompts.length === 0) return;  // Empty session — nothing to hand off
 
   // Filter prompts whose only content is workflow/control language ("继续",
@@ -73,12 +87,30 @@ export function buildAndSaveHandoff(db, sessionId, project, type, episodeSnapsho
     }
   }
 
+  // D#28 (completes D#26): observations carry the project-scoped memory_session_id, shared by
+  // parallel/sequential same-project CC sessions. Lower-bound the observation queries below to
+  // THIS CC session's start (earliest prompt epoch for ccScope) so Completed / Key Files / Key
+  // Decisions stop merging a prior session's work — the observation-side complement of
+  // working_on's cc-scoping. When ccScope is absent or its session has no prompts (MIN→null),
+  // ccWindowStart stays null and the queries run unscoped (pre-D#28 behavior). Residual: truly
+  // concurrent same-project sessions whose windows overlap can still co-attribute a few rows.
+  let ccWindowStart = null;
+  if (ccScope) {
+    const w = db.prepare(`
+      SELECT MIN(created_at_epoch) AS startEpoch FROM user_prompts
+      WHERE content_session_id = ? AND cc_session_id = ?
+    `).get(sessionId, ccScope);
+    if (typeof w?.startEpoch === 'number') ccWindowStart = w.startEpoch;
+  }
+  const obsWindowClause = ccWindowStart !== null ? 'AND created_at_epoch >= ?' : '';
+  const obsWindowParams = ccWindowStart !== null ? [ccWindowStart] : [];
+
   // 2. Completed — from observations (include narrative for richer handoff)
   const completed = db.prepare(`
     SELECT title, type, narrative FROM observations
-    WHERE memory_session_id = ? AND COALESCE(compressed_into, 0) = 0
+    WHERE memory_session_id = ? AND COALESCE(compressed_into, 0) = 0 ${obsWindowClause}
     ORDER BY created_at_epoch DESC LIMIT 15
-  `).all(sessionId);
+  `).all(sessionId, ...obsWindowParams);
 
   // 3. Recent activity — episode snapshot + full session edit history from narratives.
   // Keep only entries that represent in-flight work (file edits) or outright failures
@@ -131,9 +163,9 @@ export function buildAndSaveHandoff(db, sessionId, project, type, episodeSnapsho
   if (episodeSnapshot?.files) episodeSnapshot.files.filter(isValidFile).forEach(f => fileSet.add(f));
   const obsFiles = db.prepare(`
     SELECT files_modified FROM observations
-    WHERE memory_session_id = ? AND files_modified IS NOT NULL
+    WHERE memory_session_id = ? AND files_modified IS NOT NULL ${obsWindowClause}
     ORDER BY created_at_epoch DESC LIMIT 10
-  `).all(sessionId);
+  `).all(sessionId, ...obsWindowParams);
   for (const row of obsFiles) {
     try { JSON.parse(row.files_modified).filter(isValidFile).forEach(f => fileSet.add(f)); } catch {}
   }
@@ -142,9 +174,9 @@ export function buildAndSaveHandoff(db, sessionId, project, type, episodeSnapsho
   const decisions = db.prepare(`
     SELECT title FROM observations
     WHERE memory_session_id = ? AND COALESCE(importance, 1) >= 2
-      AND COALESCE(compressed_into, 0) = 0
+      AND COALESCE(compressed_into, 0) = 0 ${obsWindowClause}
     ORDER BY created_at_epoch DESC LIMIT 10
-  `).all(sessionId).filter(d => d.title && !LOW_SIGNAL_TITLE.test(d.title)).slice(0, 5);
+  `).all(sessionId, ...obsWindowParams).filter(d => d.title && !LOW_SIGNAL_TITLE.test(d.title)).slice(0, 5);
 
   // 6. Match keywords
   const allText = [workingOn, ...completed.map(c => c.title).filter(Boolean), unfinished].join(' ');

package/hook-llm.mjs

@@ -12,6 +12,7 @@ import {
 import { acquireLLMSlot, releaseLLMSlot } from './hook-semaphore.mjs';
 import { scrubRecord } from './lib/scrub-record.mjs';
 import { getVocabulary, computeVector } from './tfidf.mjs';
+import { DEDUP_JACCARD_THRESHOLD, AUTO_MERGE_THRESHOLD } from './lib/dedup-constants.mjs';
 import {
   RUNTIME_DIR, DEDUP_WINDOW_MS, RELATED_OBS_WINDOW_MS,
   sessionFile, getSessionId, openDb, callLLM, sleep,
@@ -148,7 +149,7 @@ export function saveObservation(obs, projectOverride, sessionIdOverride, externa
       ORDER BY created_at_epoch DESC LIMIT 10
     `).all(project, fiveMinAgo);
 
-    if (obs.title && recent.some(r => jaccardSimilarity(r.title, obs.title) > 0.7)) {
+    if (obs.title && recent.some(r => jaccardSimilarity(r.title, obs.title) > DEDUP_JACCARD_THRESHOLD)) {
       return null; // dedup: Jaccard title match
     }
 
@@ -173,8 +174,8 @@ export function saveObservation(obs, projectOverride, sessionIdOverride, externa
         WHERE project = ? AND created_at_epoch > ? AND created_at_epoch <= ?
         ORDER BY created_at_epoch DESC LIMIT 60
       `).all(project, threeDaysAgo, fiveMinAgo);
-      if (extRecent.some(r => jaccardSimilarity(r.title, obs.title) > 0.85)) {
-        return null; // dedup: low-signal Jaccard match
+      if (extRecent.some(r => jaccardSimilarity(r.title, obs.title) > AUTO_MERGE_THRESHOLD)) {
+        return null; // dedup: low-signal Jaccard match (stricter cutoff for degraded titles)
       }
     }
 

package/hook-optimize.mjs

@@ -13,6 +13,7 @@ import { callModelJSON } from './haiku-client.mjs';
 import { acquireLLMSlot, releaseLLMSlot } from './hook-semaphore.mjs';
 import { scrubRecord } from './lib/scrub-record.mjs';
 import { getVocabulary, computeVector, cosineSimilarity } from './tfidf.mjs';
+import { MERGE_JACCARD_LOW, AUTO_MERGE_THRESHOLD } from './lib/dedup-constants.mjs';
 import { DB_DIR } from './schema.mjs';
 
 const RUNTIME_DIR = join(DB_DIR, 'runtime');
@@ -331,8 +332,9 @@ export async function executeNormalize(db, force = false, { project } = {}) {
 // ─── Task 3: Cluster-merge ─────────────────────────────────────────────────
 
 const MERGE_TIME_WINDOW_MS = 30 * 86400000;
-const MERGE_JACCARD_LOW = 0.4;
-const MERGE_JACCARD_HIGH = 0.85;
+// Merge-review band [MERGE_JACCARD_LOW, AUTO_MERGE_THRESHOLD): titles in this
+// Jaccard range are LLM-reviewed for merge; at/above AUTO_MERGE_THRESHOLD they'd
+// already auto-merge elsewhere, below MERGE_JACCARD_LOW they're too dissimilar.
 
 export function findMergeCandidates(db, maxClusters = 5, { project } = {}) {
   const cutoff = Date.now() - MERGE_TIME_WINDOW_MS;
@@ -363,12 +365,14 @@ export function findMergeCandidates(db, maxClusters = 5, { project } = {}) {
       if (Math.abs(rows[i].created_at_epoch - rows[j].created_at_epoch) > MERGE_TIME_WINDOW_MS) continue;
 
       if (rows[i].minhash_sig && rows[j].minhash_sig) {
+        // 0.8 slack: the MinHash estimate is noisy, so pre-filter a band below
+        // MERGE_JACCARD_LOW rather than at it, to avoid dropping true candidates.
         const est = estimateJaccardFromMinHash(rows[i].minhash_sig, rows[j].minhash_sig);
         if (est < MERGE_JACCARD_LOW * 0.8) continue;
       }
 
       const titleSim = jaccardSimilarity(rows[i].title, rows[j].title);
-      if (titleSim >= MERGE_JACCARD_LOW && titleSim < MERGE_JACCARD_HIGH) {
+      if (titleSim >= MERGE_JACCARD_LOW && titleSim < AUTO_MERGE_THRESHOLD) {
         cluster.push(rows[j]);
         used.add(rows[j].id);
       }

package/hook-update.mjs

@@ -7,7 +7,7 @@ import { readFileSync, writeFileSync, copyFileSync, cpSync, readdirSync, existsS
 import { join, dirname } from 'node:path';
 import { pathToFileURL } from 'node:url';
 import { tmpdir, homedir } from 'node:os';
-import { DB_DIR } from './schema.mjs';
+import { DB_DIR, CODE_DIR } from './schema.mjs';
 import { debugCatch, debugLog } from './utils.mjs';
 // Local manifest is fallback only — the active manifest is loaded from the
 // extracted tarball's own source-files.mjs inside installExtractedRelease.
@@ -16,8 +16,15 @@ import { SOURCE_FILES as LOCAL_SOURCE_FILES, HOOK_SCRIPT_FILES as LOCAL_HOOK_SCR
 
 // ── Configuration ──────────────────────────────────────────
 const GITHUB_REPO = 'sdsrss/claude-mem-lite';
-const INSTALL_DIR = DB_DIR;  // ~/.claude-mem-lite/
-const STATE_FILE = join(INSTALL_DIR, 'runtime', 'update-state.json');
+// Plugin CODE location (server.mjs / package.json / install target) — always
+// homedir-rooted, NEVER follows CLAUDE_MEM_DIR (see schema.mjs CODE_DIR). Used
+// for dev-mode detection, current-version read, and the install target dir.
+const INSTALL_DIR = CODE_DIR;  // ~/.claude-mem-lite/ (code)
+// DATA/state location — runtime/update-state.json lives with the data (env-aware
+// DB_DIR), matching hook-shared RUNTIME_DIR and install.mjs doctor's read path.
+// Equal to INSTALL_DIR unless CLAUDE_MEM_DIR relocates the data dir.
+const STATE_DIR = DB_DIR;
+const STATE_FILE = join(STATE_DIR, 'runtime', 'update-state.json');
 const CHECK_INTERVAL_MS = 24 * 60 * 60 * 1000;       // 24 hours
 const FETCH_TIMEOUT_MS = 3000;                         // 3s network timeout
 const RATE_LIMIT_INTERVAL_MS = 6 * 60 * 60 * 1000;   // 6h if rate-limited
@@ -558,7 +565,7 @@ function readState() {
 
 function saveState(state) {
   try {
-    const dir = join(INSTALL_DIR, 'runtime');
+    const dir = join(STATE_DIR, 'runtime');
     mkdirSync(dir, { recursive: true });
     const tmpFile = STATE_FILE + `.tmp-${process.pid}`;
     writeFileSync(tmpFile, JSON.stringify(state, null, 2));

package/hook.mjs

@@ -63,7 +63,8 @@ import { checkForUpdate, getCachedUpdateBanner, isUpdateCheckDue } from './hook-
 import { handleLLMOptimize } from './hook-optimize.mjs';
 import { silentAutoAdopt, hasAutoAdoptMarker } from './adopt-cli.mjs';
 import { emitV270UpgradeBanner } from './lib/upgrade-banner.mjs';
-import { loadCiteBackForEpisode, buildUnsavedBugfixHint, countUnsavedBugfixShape, buildCiteRecallNudge as libBuildCiteRecallNudge, nextCiteLowStreak } from './lib/cite-back-hint.mjs';
+import { loadCiteBackForEpisode, extractCiteBackSignals, buildUnsavedBugfixHint, countUnsavedBugfixShape, buildCiteRecallNudge as libBuildCiteRecallNudge, nextCiteLowStreak } from './lib/cite-back-hint.mjs';
+import { MINHASH_PREFILTER, FUZZY_DEDUP_THRESHOLD } from './lib/dedup-constants.mjs';
 // plugin-cache-guard.mjs loaded dynamically — pre-2.31.2 installs that auto-upgraded
 // from an older hook-update.mjs SOURCE_FILES (which did not list this module) would
 // crash on static import. Degrade gracefully to no-op when the module is absent.
@@ -542,6 +543,12 @@ async function handleStop() {
           // contract test in tests/citation-tracker-userprompt.test.mjs covers it.
           try {
             const injected = extractAllInjected(transcriptPath);
+            // P5 ①: cite-back signals — observations whose warned file the agent
+            // edited this session. Union into injected so they're resolved (they
+            // were injected via pre-tool-recall) and, below, into cited so the
+            // edit promotes them even without a literal #NN in text.
+            const citeBackIds = extractCiteBackSignals(transcriptPath);
+            for (const id of citeBackIds) injected.add(id);
             if (injected.size > 0) {
               // Text-floor gate: skip decay on tool-only Stops. Without this,
               // a turn that ends on tool_use locks every injected obs as
@@ -554,6 +561,7 @@ async function handleStop() {
                 debugLog('DEBUG', 'handleStop', `citation-decay: skipped (no main-thread assistant text yet, injected=${injected.size})`);
               } else {
                 const citedMain = extractCitationsFromTranscript(transcriptPath, { mainOnly: true });
+                for (const id of citeBackIds) citedMain.add(id);
                 const r = applyCitationDecay(db, project, injected, citedMain, sessionId);
                 debugLog('DEBUG', 'handleStop', `citation-decay: touched=${r.touched} promoted=${r.promoted} demoted=${r.demoted}`);
               }
@@ -864,8 +872,6 @@ async function handleSessionStart() {
         if (!process.env.CLAUDE_MEM_SKIP_AUTO_DEDUP_FUZZY) {
           const SCAN_LIMIT = 500;
           const FUZZY_MAX_MERGES = 20;
-          const FUZZY_THRESHOLD = 0.95;
-          const MINHASH_PREFILTER = 0.7;
           const recent = db.prepare(`
             SELECT id, title, importance, created_at_epoch
             FROM observations
@@ -885,7 +891,7 @@ async function handleSessionStart() {
               for (let j = i + 1; j < recent.length; j++) {
                 if (!minhashes[j] || removed.has(recent[j].id)) continue;
                 if (estimateJaccardFromMinHash(minhashes[i], minhashes[j]) < MINHASH_PREFILTER) continue;
-                if (jaccardSimilarity(titles[i], titles[j]) < FUZZY_THRESHOLD) continue;
+                if (jaccardSimilarity(titles[i], titles[j]) < FUZZY_DEDUP_THRESHOLD) continue;
                 // Keep the higher-importance row; tiebreak by older (lower id wins access history)
                 const keep = (recent[i].importance ?? 1) >= (recent[j].importance ?? 1) ? recent[i] : recent[j];
                 const remove = keep === recent[i] ? recent[j] : recent[i];
@@ -1202,7 +1208,12 @@ async function handleUserPrompt() {
   // every downstream consumer (user_prompts INSERT, FTS query, continuation
   // detection, semantic-memory injection) sees the redacted text — single
   // source of truth for the privacy primitive.
-  const promptText = stripPrivate(rawPrompt);
+  // Strip NUL / C0 control chars (keep \t \n \r) before any downstream use: an
+  // embedded NUL terminates SQLite's C string, silently truncating the stored
+  // prompt_text at the first NUL (and breaking FTS). Single source of truth, so the
+  // user_prompts INSERT, FTS query, and continuation detection all see clean text.
+  // eslint-disable-next-line no-control-regex -- intentional: NUL/C0 strip prevents SQLite C-string truncation
+  const promptText = stripPrivate(rawPrompt).replace(/[\x00-\x08\x0b\x0c\x0e-\x1f]/g, '');
 
   const sessionId = getSessionId();
   const db = openDb();
@@ -1227,24 +1238,27 @@ async function handleUserPrompt() {
     ).get(sessionId);
     const promptNumber = bumped?.prompt_counter || 1;
 
+    // Claude Code's real session_id (CC UUID) from hook stdin. Persisted on the
+    // prompt row (cc_session_id) so buildAndSaveHandoff can scope working_on to ONE
+    // CC session — getSessionId() is project-scoped (no CC-UUID), so without this
+    // concurrent/within-TTL same-project sessions merge each other's prompts (D#26).
+    // Also scopes handoff-row injection below. Null (legacy) when stdin lacks session_id.
+    const ccSessionId = typeof hookData.session_id === 'string' && hookData.session_id.length > 0
+      ? hookData.session_id
+      : null;
+
     db.prepare(`
-      INSERT INTO user_prompts (content_session_id, prompt_text, prompt_number, created_at, created_at_epoch)
-      VALUES (?, ?, ?, ?, ?)
+      INSERT INTO user_prompts (content_session_id, prompt_text, prompt_number, cc_session_id, created_at, created_at_epoch)
+      VALUES (?, ?, ?, ?, ?, ?)
     `).run(
       sessionId,
       scrubSecrets(promptText.slice(0, 10000)),
       promptNumber,
+      ccSessionId,
       now.toISOString(), now.getTime()
     );
 
     // Cross-session handoff injection (first 3 prompts window, before semantic memory).
-    // Use Claude Code's real session_id from hook stdin to scope handoffs to this CC
-    // session — prevents cross-session bleed when running parallel sessions for the
-    // same project (see docs/bug.txt). Falls back to null (legacy behavior) if the
-    // hook input does not carry session_id.
-    const ccSessionId = typeof hookData.session_id === 'string' && hookData.session_id.length > 0
-      ? hookData.session_id
-      : null;
     if (promptNumber <= 3) {
       try {
         if (detectContinuationIntent(db, promptText, project, ccSessionId)) {