jinzd-ai-cli 0.4.88 → 0.4.90

package/dist/{chunk-YDHIU24C.js → chunk-G65IDWVP.js}

@@ -7,19 +7,23 @@ import {
 import {
   fileCheckpoints
 } from "./chunk-4BKXL7SM.js";
+import {
+  loadChatIndex,
+  searchChatMemory
+} from "./chunk-ANYYM4CF.js";
 import {
   indexProject
 } from "./chunk-NHNWUBXB.js";
 import {
   hasSemanticIndex,
   semanticSearch
-} from "./chunk-CQQQFNND.js";
+} from "./chunk-KJLJPUY2.js";
 import {
   loadIndex
 } from "./chunk-6VRJGH25.js";
 import {
   runTestsTool
-} from "./chunk-V3NMERIB.js";
+} from "./chunk-EEEAFWNK.js";
 import {
   CONFIG_DIR_NAME,
   DEFAULT_MAX_TOOL_OUTPUT_CHARS_CAP,
@@ -27,7 +31,7 @@ import {
   SUBAGENT_ALLOWED_TOOLS,
   SUBAGENT_DEFAULT_MAX_ROUNDS,
   SUBAGENT_MAX_ROUNDS_LIMIT
-} from "./chunk-VGXNE37B.js";
+} from "./chunk-WPQ4D6T3.js";
 
 // src/tools/types.ts
 function isFileWriteTool(name) {
@@ -4578,6 +4582,74 @@ ${lines.join("\n")}`;
   }
 };
 
+// src/tools/builtin/recall-memory.ts
+function formatHit(h, i) {
+  const ts = h.chunk.timestamp.slice(0, 16).replace("T", " ");
+  const title = h.chunk.sessionTitle ? ` \xB7 ${h.chunk.sessionTitle}` : "";
+  const sid = h.chunk.sessionId.slice(0, 8);
+  const score = h.score.toFixed(3);
+  const body = h.chunk.text.length > 600 ? h.chunk.text.slice(0, 600) + "\u2026" : h.chunk.text;
+  return `\u2500\u2500\u2500 Hit ${i + 1} (score ${score}, session ${sid}${title}, ${ts}) \u2500\u2500\u2500
+` + body;
+}
+var recallMemoryTool = {
+  definition: {
+    name: "recall_memory",
+    description: 'Semantic search over past chat sessions. Call this whenever the user references something that may have been discussed before ("last time", "remember", "\u4E4B\u524D", "\u4E0A\u6B21"), when context is ambiguous and continuity matters, or when you want to check what decisions or preferences have been established across prior conversations. Returns up to `topK` relevant snippets with session id, timestamp, and cosine similarity score. Prefer this over asking the user "can you remind me".',
+    parameters: {
+      query: {
+        type: "string",
+        description: "Natural-language description of what to recall. Chinese or English both work.",
+        required: true
+      },
+      topK: {
+        type: "number",
+        description: "Max number of snippets to return (default 5, max 20).",
+        required: false
+      },
+      excludeCurrentSession: {
+        type: "boolean",
+        description: "If true, exclude the current session from results (avoid echoing what you just said). Default false.",
+        required: false
+      },
+      currentSessionId: {
+        type: "string",
+        description: "Session ID to exclude when excludeCurrentSession=true. Usually the active session.",
+        required: false
+      },
+      minScore: {
+        type: "number",
+        description: "Drop hits below this cosine score. Default 0.25. Raise to 0.35+ for stricter matches.",
+        required: false
+      }
+    },
+    dangerous: false
+  },
+  async execute(args) {
+    const query = String(args["query"] ?? "").trim();
+    if (!query) throw new ToolError("recall_memory", "query is required");
+    const topK = Math.max(1, Math.min(20, Number(args["topK"] ?? 5)));
+    const excludeCurrent = Boolean(args["excludeCurrentSession"]);
+    const currentId = args["currentSessionId"] ? String(args["currentSessionId"]) : void 0;
+    const minScore = args["minScore"] !== void 0 ? Number(args["minScore"]) : 0.25;
+    const status = loadChatIndex();
+    if (!status) {
+      return "No chat memory index found. The index is built on REPL startup, or run `/memory rebuild` manually. If you have no past sessions yet, this is expected.";
+    }
+    const hits = await searchChatMemory(query, {
+      topK,
+      minScore,
+      excludeSessionId: excludeCurrent ? currentId : void 0
+    });
+    if (hits.length === 0) {
+      return `No memories matched "${query}" above score ${minScore}. Index has ${status.idx.chunks.length} chunks across ${Object.keys(status.idx.sessionMtimes).length} sessions. Consider lowering minScore to 0.15 or rephrasing the query.`;
+    }
+    const header = `Found ${hits.length} memory hit(s) for "${query}" (min-score ${minScore}):
+`;
+    return header + "\n" + hits.map(formatHit).join("\n\n");
+  }
+};
+
 // src/core/token-estimator.ts
 var CJK_REGEX = /[\u2E80-\u9FFF\uA000-\uA4FF\uAC00-\uD7FF\uF900-\uFAFF\uFE30-\uFE4F\uFF00-\uFFEF]/g;
 function estimateTokens(text) {
@@ -4637,6 +4709,7 @@ var ToolRegistry = class {
     this.register(getOutlineTool);
     this.register(findReferencesTool);
     this.register(searchCodeTool);
+    this.register(recallMemoryTool);
   }
   register(tool) {
     this.tools.set(tool.definition.name, tool);

package/dist/chunk-JV5N65KN.js

@@ -0,0 +1,50 @@
+// src/symbols/embedder.ts
+import path from "path";
+import os from "os";
+import fs from "fs";
+var EMBEDDING_MODEL_ID = "Xenova/paraphrase-multilingual-MiniLM-L12-v2";
+var EMBEDDING_DIM = 384;
+var pipelinePromise = null;
+function cacheDir() {
+  return path.join(os.homedir(), ".aicli", "models");
+}
+async function getEmbedder() {
+  if (pipelinePromise) return pipelinePromise;
+  pipelinePromise = (async () => {
+    const mod = await import("@huggingface/transformers");
+    const dir = cacheDir();
+    fs.mkdirSync(dir, { recursive: true });
+    mod.env.cacheDir = dir;
+    mod.env.allowRemoteModels = true;
+    mod.env.allowLocalModels = true;
+    const pipe = await mod.pipeline("feature-extraction", EMBEDDING_MODEL_ID, {
+      // Keep the ONNX session in float32; int8 quantization exists but the
+      // quality drop on short code identifiers is noticeable.
+      dtype: "fp32"
+    });
+    return pipe;
+  })();
+  return pipelinePromise;
+}
+async function embed(texts) {
+  if (texts.length === 0) return [];
+  const pipe = await getEmbedder();
+  const out = await pipe(texts, { pooling: "mean", normalize: true });
+  const batch = texts.length;
+  const dim = EMBEDDING_DIM;
+  const rows = new Array(batch);
+  for (let i = 0; i < batch; i++) {
+    rows[i] = new Float32Array(out.data.buffer, out.data.byteOffset + i * dim * 4, dim).slice();
+  }
+  return rows;
+}
+async function embedOne(text) {
+  const [vec] = await embed([text]);
+  return vec;
+}
+
+export {
+  EMBEDDING_DIM,
+  embed,
+  embedOne
+};

package/dist/chunk-KHYD3WXE.js

@@ -0,0 +1,52 @@
+#!/usr/bin/env node
+
+// src/symbols/embedder.ts
+import path from "path";
+import os from "os";
+import fs from "fs";
+var EMBEDDING_MODEL_ID = "Xenova/paraphrase-multilingual-MiniLM-L12-v2";
+var EMBEDDING_DIM = 384;
+var pipelinePromise = null;
+function cacheDir() {
+  return path.join(os.homedir(), ".aicli", "models");
+}
+async function getEmbedder() {
+  if (pipelinePromise) return pipelinePromise;
+  pipelinePromise = (async () => {
+    const mod = await import("@huggingface/transformers");
+    const dir = cacheDir();
+    fs.mkdirSync(dir, { recursive: true });
+    mod.env.cacheDir = dir;
+    mod.env.allowRemoteModels = true;
+    mod.env.allowLocalModels = true;
+    const pipe = await mod.pipeline("feature-extraction", EMBEDDING_MODEL_ID, {
+      // Keep the ONNX session in float32; int8 quantization exists but the
+      // quality drop on short code identifiers is noticeable.
+      dtype: "fp32"
+    });
+    return pipe;
+  })();
+  return pipelinePromise;
+}
+async function embed(texts) {
+  if (texts.length === 0) return [];
+  const pipe = await getEmbedder();
+  const out = await pipe(texts, { pooling: "mean", normalize: true });
+  const batch = texts.length;
+  const dim = EMBEDDING_DIM;
+  const rows = new Array(batch);
+  for (let i = 0; i < batch; i++) {
+    rows[i] = new Float32Array(out.data.buffer, out.data.byteOffset + i * dim * 4, dim).slice();
+  }
+  return rows;
+}
+async function embedOne(text) {
+  const [vec] = await embed([text]);
+  return vec;
+}
+
+export {
+  EMBEDDING_DIM,
+  embed,
+  embedOne
+};

package/dist/{chunk-CQQQFNND.js → chunk-KJLJPUY2.js}

@@ -3,13 +3,15 @@ import {
   loadIndex
 } from "./chunk-6VRJGH25.js";
 import {
-  EMBEDDING_DIM,
-  embed,
-  embedOne,
   loadVectorStore,
   saveVectorStore,
   searchVectorStore
-} from "./chunk-PFYAAX2S.js";
+} from "./chunk-2DXY7UGF.js";
+import {
+  EMBEDDING_DIM,
+  embed,
+  embedOne
+} from "./chunk-KHYD3WXE.js";
 
 // src/symbols/semantic.ts
 function pathTokens(absFile, root) {

package/dist/{chunk-GTKJUEBS.js → chunk-MO7MWNWC.js}

@@ -2,13 +2,15 @@ import {
   loadIndex
 } from "./chunk-BJAT4GNC.js";
 import {
-  EMBEDDING_DIM,
-  embed,
-  embedOne,
   loadVectorStore,
   saveVectorStore,
   searchVectorStore
-} from "./chunk-XMA222FQ.js";
+} from "./chunk-PASCDYMH.js";
+import {
+  EMBEDDING_DIM,
+  embed,
+  embedOne
+} from "./chunk-JV5N65KN.js";
 
 // src/symbols/semantic.ts
 function pathTokens(absFile, root) {

package/dist/{chunk-XMA222FQ.js → chunk-PASCDYMH.js}

@@ -1,70 +1,27 @@
-// src/symbols/vector-store.ts
-import fs2 from "fs";
-import path2 from "path";
-import os2 from "os";
-import crypto from "crypto";
+import {
+  EMBEDDING_DIM
+} from "./chunk-JV5N65KN.js";
 
-// src/symbols/embedder.ts
+// src/symbols/vector-store.ts
+import fs from "fs";
 import path from "path";
 import os from "os";
-import fs from "fs";
-var EMBEDDING_MODEL_ID = "Xenova/paraphrase-multilingual-MiniLM-L12-v2";
-var EMBEDDING_DIM = 384;
-var pipelinePromise = null;
-function cacheDir() {
-  return path.join(os.homedir(), ".aicli", "models");
-}
-async function getEmbedder() {
-  if (pipelinePromise) return pipelinePromise;
-  pipelinePromise = (async () => {
-    const mod = await import("@huggingface/transformers");
-    const dir = cacheDir();
-    fs.mkdirSync(dir, { recursive: true });
-    mod.env.cacheDir = dir;
-    mod.env.allowRemoteModels = true;
-    mod.env.allowLocalModels = true;
-    const pipe = await mod.pipeline("feature-extraction", EMBEDDING_MODEL_ID, {
-      // Keep the ONNX session in float32; int8 quantization exists but the
-      // quality drop on short code identifiers is noticeable.
-      dtype: "fp32"
-    });
-    return pipe;
-  })();
-  return pipelinePromise;
-}
-async function embed(texts) {
-  if (texts.length === 0) return [];
-  const pipe = await getEmbedder();
-  const out = await pipe(texts, { pooling: "mean", normalize: true });
-  const batch = texts.length;
-  const dim = EMBEDDING_DIM;
-  const rows = new Array(batch);
-  for (let i = 0; i < batch; i++) {
-    rows[i] = new Float32Array(out.data.buffer, out.data.byteOffset + i * dim * 4, dim).slice();
-  }
-  return rows;
-}
-async function embedOne(text) {
-  const [vec] = await embed([text]);
-  return vec;
-}
-
-// src/symbols/vector-store.ts
+import crypto from "crypto";
 var MAGIC = 1094927190;
 var VERSION = 1;
 var HEADER_BYTES = 16;
 function indexDir() {
-  return path2.join(os2.homedir(), ".aicli", "index");
+  return path.join(os.homedir(), ".aicli", "index");
 }
 function projectHash(root) {
-  return crypto.createHash("sha1").update(path2.resolve(root).toLowerCase()).digest("hex").slice(0, 16);
+  return crypto.createHash("sha1").update(path.resolve(root).toLowerCase()).digest("hex").slice(0, 16);
 }
 function vecPath(root) {
-  return path2.join(indexDir(), `${projectHash(root)}.vec`);
+  return path.join(indexDir(), `${projectHash(root)}.vec`);
 }
 function emptyVectorStore(root) {
   return {
-    root: path2.resolve(root),
+    root: path.resolve(root),
     count: 0,
     dim: EMBEDDING_DIM,
     vectors: new Float32Array(0),
@@ -79,7 +36,7 @@ function saveVectorStore(root, indices, vectors) {
   }
   const count = indices.length;
   const dir = indexDir();
-  fs2.mkdirSync(dir, { recursive: true });
+  fs.mkdirSync(dir, { recursive: true });
   const totalBytes = HEADER_BYTES + count * 4 + count * EMBEDDING_DIM * 4;
   const buf = Buffer.alloc(totalBytes);
   buf.writeUInt32LE(MAGIC, 0);
@@ -90,15 +47,15 @@ function saveVectorStore(root, indices, vectors) {
   Buffer.from(vectors.buffer, vectors.byteOffset, vectors.byteLength).copy(buf, HEADER_BYTES + count * 4);
   const target = vecPath(root);
   const tmp = `${target}.tmp`;
-  fs2.writeFileSync(tmp, buf);
-  fs2.renameSync(tmp, target);
+  fs.writeFileSync(tmp, buf);
+  fs.renameSync(tmp, target);
 }
 function loadVectorStore(root) {
   const p = vecPath(root);
-  if (!fs2.existsSync(p)) return null;
+  if (!fs.existsSync(p)) return null;
   let buf;
   try {
-    buf = fs2.readFileSync(p);
+    buf = fs.readFileSync(p);
   } catch {
     return null;
   }
@@ -119,12 +76,12 @@ function loadVectorStore(root) {
       buf.byteOffset + HEADER_BYTES + count * 4 + count * dim * 4
     )
   );
-  return { root: path2.resolve(root), count, dim, vectors, symbolIdx };
+  return { root: path.resolve(root), count, dim, vectors, symbolIdx };
 }
 function clearVectorStore(root) {
   const p = vecPath(root);
   try {
-    if (fs2.existsSync(p)) fs2.unlinkSync(p);
+    if (fs.existsSync(p)) fs.unlinkSync(p);
   } catch {
   }
 }
@@ -156,9 +113,6 @@ function searchVectorStore(store, queryVec, k) {
 }
 
 export {
-  EMBEDDING_DIM,
-  embed,
-  embedOne,
   emptyVectorStore,
   saveVectorStore,
   loadVectorStore,

package/dist/{chunk-VGXNE37B.js → chunk-WPQ4D6T3.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 
 // src/core/constants.ts
-var VERSION = "0.4.88";
+var VERSION = "0.4.90";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";