mcp-probe-kit 3.2.0 → 3.6.0

package/README.md

@@ -15,7 +15,7 @@
 
 > **Talk is cheap, show me the Context.**
 > 
-> mcp-probe-kit is a protocol-level toolkit designed for developers who want AI to truly understand their project's intent. It's not just a collection of 27 tools—it's a context-aware system that helps AI agents grasp what you're building.
+> mcp-probe-kit is a protocol-level toolkit designed for developers who want AI to truly understand their project's intent. It's not just a collection of 30 tools—it's a context-aware system that helps AI agents grasp what you're building.
 
 **Languages**: [English](README.md) | [简体中文](i18n/README.zh-CN.md) | [日本語](i18n/README.ja-JP.md) | [한국어](i18n/README.ko-KR.md) | [Español](i18n/README.es-ES.md) | [Français](i18n/README.fr-FR.md) | [Deutsch](i18n/README.de-DE.md) | [Português (BR)](i18n/README.pt-BR.md)
 
@@ -26,7 +26,7 @@
 
 > 🚀 AI-Powered Complete Development Toolkit - Covering the Entire Development Lifecycle
 
-A powerful MCP (Model Context Protocol) server providing **27 tools** covering the complete workflow from product analysis to final release (Requirements → Design → Development → Quality → Release), all tools support **structured output**.
+A powerful MCP (Model Context Protocol) server providing **30 tools** covering the complete workflow from product analysis to final release (Requirements → Design → Development → Quality → Release), all tools support **structured output**.
 
 **🎉 v3.0 Major Update**: Streamlined tool count, focus on core competencies, eliminate choice paralysis, let AI do more native work
 
@@ -42,7 +42,7 @@ A powerful MCP (Model Context Protocol) server providing **27 tools** covering t
 
 - [Quick Start](https://mcp-probe-kit.bytezonex.com/pages/getting-started.html) - Setup in 5 minutes
 - [Local Memory Stack (Qdrant + Nomic Embed)](docs/memory-local-setup.md) - Docker Compose, ports `50008` / `50012`, MCP env
-- [All Tools](https://mcp-probe-kit.bytezonex.com/pages/all-tools.html) - Complete list of 27 tools
+- [All Tools](https://mcp-probe-kit.bytezonex.com/pages/all-tools.html) - Complete list of 29 tools
 - [Best Practices](https://mcp-probe-kit.bytezonex.com/pages/examples.html) - Full development workflow guide
 - [v3.0 Migration Guide](https://mcp-probe-kit.bytezonex.com/pages/migration.html) - Upgrade from v2.x to v3.0
 
@@ -50,7 +50,7 @@ A powerful MCP (Model Context Protocol) server providing **27 tools** covering t
 
 ## ✨ Core Features
 
-### 📦 27 Tools
+### 📦 29 Tools
 
 - **🔄 Workflow Orchestration** (6 tools) - One-click complex development workflows
   - `start_feature`, `start_bugfix`, `start_onboard`, `start_ui`, `start_product`, `start_ralph`
@@ -64,8 +64,18 @@ A powerful MCP (Model Context Protocol) server providing **27 tools** covering t
   - `init_project`, `init_project_context`, `add_feature`, `check_spec`, `estimate`, `interview`, `ask_user`
 - **🎨 UI/UX Utilities** (3 tools) - Design systems and UI data synchronization
   - `ui_design_system`, `ui_search`, `sync_ui_data`
-- **🧠 Memory** (4 tools) - Reusable asset memory
-  - `search_memory`, `read_memory_asset`, `memorize_asset`, `scan_and_extract_patterns`
+- **🧠 Memory** (6 tools) - Reusable asset memory
+  - `search_memory`, `read_memory_asset`, `memorize_asset`, `update_memory_asset`, `delete_memory_asset`, `scan_and_extract_patterns`
+
+### 🛡️ Quality Constraints (single source of truth)
+
+All hard quality rules live in one module (`src/lib/quality-constraints.ts`) and are injected into `code_review`, the `add_feature` task templates, and the UI tools. Change once, apply everywhere — inspired by [taste-skill](https://github.com/Leonxlnx/taste-skill) and [impeccable](https://github.com/pbakaus/impeccable).
+
+- **Code limits**: single file ≤ 500 lines (split into modules/components when exceeded), function ≤ 50 lines, nesting ≤ 4, parameters ≤ 3.
+- **Completeness blacklist**: `code_review` flags placeholder/elision patterns (`// ...`, `// TODO`, `// rest of code`, bare `...`) as CRITICAL — "a partial output is a broken output".
+- **Anti-laziness task templates**: `add_feature` tasks now carry a Scope-lock deliverable count, a mandatory evidence block (read code before writing), a per-file line budget, and a binary zero-tolerance rule for placeholders. `check_spec` validates these (missing Scope-lock = error, thin task without evidence = warning).
+- **UI hard red lines**: numeric, machine-checkable rules — 4pt spacing scale, WCAG contrast (4.5/3/3), type scale ≥ 1.25, hero font ≤ 6rem, OKLCH, eight interaction states, cognitive load ≤ 4, motion 150-300ms.
+- **UI banned list + Pre-Flight checklist**: match-and-refuse blacklist for AI slop (default Inter/Roboto, AI purple-blue gradients, gradient text, cookie-cutter card grids, em-dash, cream/beige body backgrounds, nested cards) plus a delivery-gate self-check matrix.
 
 ### 🧠 Code Graph Bridge (GitNexus)
 
@@ -98,6 +108,8 @@ A powerful MCP (Model Context Protocol) server providing **27 tools** covering t
 - `search_memory` - Semantic search across the shared memory pool (optionally prefer `type` / `tags`); text output includes `id`, `score`, summary, description, and a `--- content ---` body (default up to 1500 chars via `MEMORY_SEARCH_CONTENT_MAX_CHARS`)
 - `memorize_asset` - Persist reusable code/spec/pattern assets into vector memory
 - `read_memory_asset` - Read full asset content by `asset_id` (text output includes the full `content` body)
+- `update_memory_asset` - Update an existing asset by `asset_id` (preserves ID; `content` changes re-embed)
+- `delete_memory_asset` - Delete an asset by `asset_id` from the shared pool
 - `scan_and_extract_patterns` - Extract reusable patterns from code/file/directory before deciding whether to persist
 
 **Cross-repo memory pools:** do not rely on `source_project` / `source_path` for shared retrieval; put file paths in `content` instead. Search injection hides foreign `sourcePath` unless `MEMORY_REPO_ID` matches or `MEMORY_SEARCH_SHOW_SOURCE=true`.
@@ -468,7 +480,13 @@ Use in config file:
 
 ### Optional Memory System Setup
 
-If you want to use `memorize_asset`, `read_memory_asset`, and `scan_and_extract_patterns`, you need both:
+If you want to use `memorize_asset`, `update_memory_asset`, `read_memory_asset`, `delete_memory_asset`, and `scan_and_extract_patterns`, configure as follows:
+
+- **Qdrant only** (`MEMORY_QDRANT_URL`): `read_memory_asset`, `delete_memory_asset`
+- **Qdrant + embedding** (all three `MEMORY_*` write/search vars): `search_memory`, `memorize_asset`, `update_memory_asset`
+- **No memory backend**: `scan_and_extract_patterns` (local scan only; persist via `memorize_asset` when ready)
+
+For full write/search you need both:
 
 1. A **Qdrant** vector database
 2. An **embedding service** in either `ollama` or `openai-compatible` mode

package/build/index.js

@@ -5,11 +5,16 @@ import { InMemoryTaskMessageQueue, InMemoryTaskStore, } from "@modelcontextproto
 import { CallToolRequestSchema, ListToolsRequestSchema, ListResourcesRequestSchema, ProgressNotificationSchema, ReadResourceRequestSchema, } from "@modelcontextprotocol/sdk/types.js";
 import * as fs from "node:fs";
 import * as path from "node:path";
-import { initProject, gencommit, codeReview, codeInsight, gentest, refactor, initProjectContext, addFeature, fixBug, estimate, checkSpec, startFeature, startBugfix, startOnboard, startRalph, interview, askUser, uiDesignSystem, uiSearch, syncUiData, startUi, startProduct, gitWorkReport, searchMemory, readMemoryAsset, memorizeAsset, scanAndExtractPatterns } from "./tools/index.js";
+import { initProject, gencommit, codeReview, codeInsight, gentest, refactor, initProjectContext, addFeature, fixBug, estimate, checkSpec, workflow, startFeature, startBugfix, startOnboard, startRalph, interview, askUser, uiDesignSystem, uiSearch, syncUiData, startUi, startProduct, gitWorkReport, searchMemory, readMemoryAsset, memorizeAsset, deleteMemoryAsset, updateMemoryAsset, scanAndExtractPatterns } from "./tools/index.js";
 import { VERSION, NAME } from "./version.js";
 import { allToolSchemas } from "./schemas/index.js";
 import { filterTools, getToolsetFromEnv } from "./lib/toolset-manager.js";
 import { withToolAnnotations } from "./lib/tool-annotations.js";
+import { withOutputSchema } from "./lib/output-schema-registry.js";
+import { shouldAutoEscalateToTask } from "./lib/task-defaults.js";
+import { attachHandles } from "./lib/handles.js";
+import { buildMcpAppHtml, isMcpUiAppTool } from "./lib/mcp-apps.js";
+import { ensureMcpProbeKitBootstrapForToolCall } from "./lib/workflow-skill-installer.js";
 import { isAbortError, } from "./lib/tool-execution-context.js";
 const EXTENSIONS_CAPABILITY_KEY = "io.github.mybolide/extensions";
 const MAX_UI_APP_RESOURCES = 30;
@@ -82,80 +87,6 @@ function withTraceMeta(result, traceMeta) {
         },
     };
 }
-function escapeHtml(value) {
-    return value
-        .replace(/&/g, "&")
-        .replace(/</g, "&lt;")
-        .replace(/>/g, "&gt;")
-        .replace(/"/g, "&quot;")
-        .replace(/'/g, "&#39;");
-}
-function isUiTool(name) {
-    return [
-        "ui_design_system",
-        "ui_search",
-        "sync_ui_data",
-        "start_ui",
-        "start_product",
-    ].includes(name);
-}
-function buildUiResourceHtml(name, args, result) {
-    const structured = result.structuredContent
-        ? JSON.stringify(result.structuredContent, null, 2)
-        : "{}";
-    const argJson = JSON.stringify(args ?? {}, null, 2);
-    const textBlocks = Array.isArray(result.content)
-        ? result.content
-            .map((item) => {
-            if (!item || typeof item !== "object") {
-                return "";
-            }
-            const text = item.text;
-            return typeof text === "string" ? text : "";
-        })
-            .filter(Boolean)
-            .join("\n\n")
-        : "";
-    const now = new Date().toISOString();
-    return `<!doctype html>
-<html lang="zh-CN">
-<head>
-  <meta charset="utf-8">
-  <meta name="viewport" content="width=device-width, initial-scale=1">
-  <title>${escapeHtml(name)} · MCP Apps</title>
-  <style>
-    :root { color-scheme: light; }
-    body { font-family: "Segoe UI", "PingFang SC", "Microsoft YaHei", sans-serif; margin: 0; background: #f4f7fb; color: #1e2a35; }
-    .wrap { max-width: 960px; margin: 0 auto; padding: 24px; }
-    .card { background: #fff; border-radius: 14px; padding: 18px; box-shadow: 0 4px 18px rgba(30,42,53,.08); margin-bottom: 16px; }
-    h1 { margin: 0 0 8px; font-size: 24px; }
-    h2 { margin: 0 0 10px; font-size: 16px; color: #2f4a65; }
-    pre { white-space: pre-wrap; word-break: break-word; background: #0f1720; color: #d9e7f7; border-radius: 10px; padding: 12px; font-size: 12px; line-height: 1.45; }
-    .meta { color: #4f6880; font-size: 12px; }
-  </style>
-</head>
-<body>
-  <div class="wrap">
-    <div class="card">
-      <h1>${escapeHtml(name)}</h1>
-      <div class="meta">Generated at ${escapeHtml(now)} · MCP Apps preview</div>
-    </div>
-    <div class="card">
-      <h2>Text Output</h2>
-      <pre>${escapeHtml(textBlocks || "(no text output)")}</pre>
-    </div>
-    <div class="card">
-      <h2>Structured Content</h2>
-      <pre>${escapeHtml(structured)}</pre>
-    </div>
-    <div class="card">
-      <h2>Arguments</h2>
-      <pre>${escapeHtml(argJson)}</pre>
-    </div>
-  </div>
-</body>
-</html>`;
-}
 function putUiAppResource(toolName, args, result) {
     const uid = `${Date.now()}-${Math.random().toString(16).slice(2, 10)}`;
     const uri = `ui://mcp-probe-kit/${toolName}/${uid}`;
@@ -164,7 +95,7 @@ function putUiAppResource(toolName, args, result) {
         name: `UI Preview · ${toolName}`,
         description: `MCP Apps preview generated by ${toolName}`,
         mimeType: "text/html",
-        text: buildUiResourceHtml(toolName, args, result),
+        text: buildMcpAppHtml(toolName, args, result),
         createdAt: new Date().toISOString(),
     };
     uiAppResources.set(uri, entry);
@@ -384,13 +315,26 @@ function rememberGraphSnapshot(toolName, result) {
     }
     return snapshot;
 }
+function withStructuredHandles(result, handles) {
+    if (!result.structuredContent || typeof result.structuredContent !== "object" || Array.isArray(result.structuredContent)) {
+        return result;
+    }
+    return {
+        ...result,
+        structuredContent: attachHandles(result.structuredContent, handles),
+    };
+}
 function decorateResult(toolName, args, raw, traceMeta) {
     let result = withTraceMeta(raw, traceMeta);
     const snapshot = rememberGraphSnapshot(toolName, result);
     if (snapshot) {
         result = withGraphSnapshotMeta(result, snapshot);
+        result = withStructuredHandles(result, {
+            graph_snapshot: snapshot.uri,
+            graph_resource: snapshot.uri,
+        });
     }
-    if (uiAppsEnabled && isUiTool(toolName) && !result.isError) {
+    if (uiAppsEnabled && isMcpUiAppTool(toolName) && !result.isError) {
         const resourceUri = putUiAppResource(toolName, args, result);
         result = withUiResourceMeta(result, resourceUri);
     }
@@ -411,10 +355,23 @@ server.setRequestHandler(ListToolsRequestSchema, async () => {
     const filteredTools = filterTools(allToolSchemas, toolset);
     console.error(`[MCP Probe Kit] 当前工具集: ${toolset} (${filteredTools.length}/${allToolSchemas.length} 个工具)`);
     return {
-        tools: filteredTools.map(withToolAnnotations),
+        tools: filteredTools.map((tool) => withOutputSchema(withToolAnnotations(tool))),
     };
 });
 async function executeTool(name, args, context) {
+    const bootstrap = ensureMcpProbeKitBootstrapForToolCall(name, args);
+    if (bootstrap?.skill.created) {
+        console.error(`[MCP Probe Kit] 已创建 MCP Skill: ${bootstrap.skill.skillRelPath} (v${bootstrap.skill.version})`);
+    }
+    else if (bootstrap?.skill.updated) {
+        console.error(`[MCP Probe Kit] 已升级 MCP Skill: ${bootstrap.skill.skillRelPath} ${bootstrap.skill.previousVersion ?? "?"} → v${bootstrap.skill.version}`);
+    }
+    if (bootstrap?.agentsMd.created) {
+        console.error(`[MCP Probe Kit] 已创建 AGENTS.md（含 Skill 引用）: ${bootstrap.agentsMd.path}`);
+    }
+    else if (bootstrap?.agentsMd.updated) {
+        console.error(`[MCP Probe Kit] 已更新 AGENTS.md（添加 Skill 引用）: ${bootstrap.agentsMd.path}`);
+    }
     switch (name) {
         case "init_project":
             return await initProject(args);
@@ -430,6 +387,8 @@ async function executeTool(name, args, context) {
             return await refactor(args);
         case "init_project_context":
             return await initProjectContext(args);
+        case "workflow":
+            return await workflow(args);
         case "add_feature":
             return await addFeature(args);
         case "check_spec":
@@ -468,6 +427,10 @@ async function executeTool(name, args, context) {
             return await readMemoryAsset(args);
         case "memorize_asset":
             return await memorizeAsset(args);
+        case "delete_memory_asset":
+            return await deleteMemoryAsset(args);
+        case "update_memory_asset":
+            return await updateMemoryAsset(args);
         case "scan_and_extract_patterns":
             return await scanAndExtractPatterns(args);
         default:
@@ -491,8 +454,11 @@ function isTerminalTaskStatus(status) {
 // 处理工具调用
 server.setRequestHandler(CallToolRequestSchema, async (request, extra) => {
     const { name, arguments: args } = request.params;
-    const taskRequest = request.params.task;
+    let taskRequest = request.params.task;
     const traceMeta = getTraceMeta(extra._meta);
+    if (shouldAutoEscalateToTask(name, Boolean(taskRequest))) {
+        taskRequest = taskRequest ?? {};
+    }
     const emitProgress = async (progress, message) => {
         const progressToken = extra._meta?.progressToken;
         if (progressToken === undefined) {

package/build/lib/__tests__/agents-md-template.unit.test.js

@@ -20,6 +20,7 @@ describe('generateAgentsMdInner', () => {
         expect(md).toContain('记忆');
         expect(md).toContain('start_bugfix');
         expect(md).toContain('memorize_asset');
+        expect(md).toContain('mcp-probe-kit/SKILL.md');
         expect(md).toContain('bugfix');
         expect(md).toContain('search_memory');
         expect(md).toContain('自动注入');

package/build/lib/__tests__/dev-workflow.unit.test.js

@@ -0,0 +1,22 @@
+import { describe, expect, test } from 'vitest';
+import { buildDevWorkflow, detectWorkflowScenario } from '../dev-workflow.js';
+describe('dev-workflow', () => {
+    test('识别 bugfix 场景', () => {
+        const result = detectWorkflowScenario('登录报错 NullReference 需要排查修复');
+        expect(result.scenario).toBe('bugfix');
+    });
+    test('识别 feature 场景', () => {
+        const result = detectWorkflowScenario('开发用户认证新功能');
+        expect(result.scenario).toBe('feature');
+    });
+    test('bugfix 计划首工具为 start_bugfix', () => {
+        const plan = buildDevWorkflow('TypeError in checkout');
+        expect(plan.firstTool).toBe('start_bugfix');
+        expect(plan.phases.some((p) => p.steps.some((s) => s.tool === 'start_bugfix'))).toBe(true);
+    });
+    test('显式 scenario 覆盖推断', () => {
+        const plan = buildDevWorkflow('随便', { scenario: 'ui' });
+        expect(plan.scenario).toBe('ui');
+        expect(plan.firstTool).toBe('start_ui');
+    });
+});

package/build/lib/__tests__/handles.unit.test.js

@@ -0,0 +1,19 @@
+import { describe, expect, test } from 'vitest';
+import { attachHandles, buildMemoryAssetHandles, mergeHandles, } from '../../lib/handles.js';
+describe('handles', () => {
+    test('buildMemoryAssetHandles 生成 read 默认工具名', () => {
+        const handles = buildMemoryAssetHandles([{ id: 'a-1', name: 'demo' }]);
+        expect(handles).toEqual([{ id: 'a-1', name: 'demo', tool: 'read_memory_asset' }]);
+    });
+    test('attachHandles 合并已有 handles', () => {
+        const result = attachHandles({ handles: { graph_resource: 'probe://graph/latest' } }, { memory_assets: buildMemoryAssetHandles([{ id: 'x' }]) });
+        expect(result.handles.graph_resource).toBe('probe://graph/latest');
+        expect(result.handles.memory_assets).toHaveLength(1);
+    });
+    test('mergeHandles 后者覆盖 graph 字段', () => {
+        expect(mergeHandles({ graph_resource: 'a' }, { graph_resource: 'b', graph_snapshot: 'probe://graph/s1' })).toEqual({
+            graph_resource: 'b',
+            graph_snapshot: 'probe://graph/s1',
+        });
+    });
+});

package/build/lib/__tests__/mcp-apps.unit.test.d.ts

@@ -0,0 +1 @@
+export {};

package/build/lib/__tests__/mcp-apps.unit.test.js

@@ -0,0 +1,52 @@
+import { describe, expect, test } from 'vitest';
+import { buildCodeInsightAppHtml, buildSearchMemoryAppHtml, isMcpUiAppTool, } from '../mcp-apps.js';
+describe('mcp-apps', () => {
+    test('search_memory 与 code_insight 属于 UI App 工具', () => {
+        expect(isMcpUiAppTool('search_memory')).toBe(true);
+        expect(isMcpUiAppTool('code_insight')).toBe(true);
+        expect(isMcpUiAppTool('gencommit')).toBe(false);
+    });
+    test('search_memory App 渲染命中卡片与 handles', () => {
+        const html = buildSearchMemoryAppHtml({ query: 'proxy bug' }, {
+            structuredContent: {
+                query: 'proxy bug',
+                count: 1,
+                results: [
+                    {
+                        id: 'asset-1',
+                        name: 'feishu-proxy',
+                        type: 'bugfix',
+                        score: 0.88,
+                        summary: 'proxy mismatch',
+                        content: '【修复】done',
+                        tags: ['bugfix'],
+                    },
+                ],
+                handles: {
+                    memory_assets: [{ id: 'asset-1', tool: 'read_memory_asset' }],
+                },
+            },
+        });
+        expect(html).toContain('search_memory');
+        expect(html).toContain('feishu-proxy');
+        expect(html).toContain('asset-1');
+        expect(html).toContain('read_memory_asset');
+        expect(html).toContain('proxy mismatch');
+    });
+    test('code_insight App 渲染状态与执行表', () => {
+        const html = buildCodeInsightAppHtml({ mode: 'auto' }, {
+            structuredContent: {
+                status: 'ok',
+                provider: 'gitnexus',
+                summary: 'login flow analyzed',
+                mode: { requested: 'auto', resolved: 'query' },
+                executions: [{ tool: 'query', ok: true, text: 'found login handler' }],
+                handles: { graph_resource: 'probe://graph/latest' },
+            },
+        });
+        expect(html).toContain('code_insight');
+        expect(html).toContain('login flow analyzed');
+        expect(html).toContain('probe://graph/latest');
+        expect(html).toContain('found login handler');
+    });
+});

package/build/lib/__tests__/mcp-tool-skill-registry.unit.test.d.ts

@@ -0,0 +1 @@
+export {};

package/build/lib/__tests__/mcp-tool-skill-registry.unit.test.js

@@ -0,0 +1,18 @@
+import { describe, expect, it } from "vitest";
+import { allToolSchemas } from "../../schemas/index.js";
+import { formatSkillRegistryMismatchMessage, listMcpToolSkillRegistryNames, validateMcpToolSkillRegistry, } from "../mcp-tool-skill-registry.js";
+import { generateWorkflowSkillContent } from "../workflow-skill-template.js";
+describe("mcp-tool-skill-registry", () => {
+    it("与 allToolSchemas 工具名一一对应", () => {
+        const registered = allToolSchemas.map((tool) => tool.name).sort();
+        const result = validateMcpToolSkillRegistry(registered);
+        expect(result.ok, formatSkillRegistryMismatchMessage(result)).toBe(true);
+        expect(listMcpToolSkillRegistryNames().length).toBe(registered.length);
+    });
+    it("生成的 Skill 正文包含每个注册工具", () => {
+        const content = generateWorkflowSkillContent("0.0.0-test");
+        for (const name of listMcpToolSkillRegistryNames()) {
+            expect(content).toContain(`\`${name}\``);
+        }
+    });
+});

package/build/lib/__tests__/memory-client.unit.test.js

@@ -80,4 +80,109 @@ describe('memory-client 去重逻辑', () => {
         expect(result.normalizedContentHash).toBe(existingAsset.normalizedContentHash);
         expect(fetchMock.mock.calls.some(([url]) => String(url).includes('/points?wait=true'))).toBe(false);
     });
+    test('deleteAsset 删除已存在资产并返回快照', async () => {
+        vi.stubEnv('MEMORY_QDRANT_URL', 'http://127.0.0.1:50008');
+        const asset = {
+            id: 'asset-to-delete',
+            name: 'ObsoletePattern',
+            type: 'pattern',
+            description: '过时模式',
+            summary: '不再使用',
+            content: 'old snippet',
+            tags: ['pattern'],
+            confidence: 0.7,
+            createdAt: '2026-01-01T00:00:00.000Z',
+            updatedAt: '2026-01-01T00:00:00.000Z',
+        };
+        const fetchMock = vi.fn(async (input, init) => {
+            const url = String(input);
+            if (url.includes(`/points/${asset.id}`) && init?.method !== 'POST' && init?.method !== 'PUT') {
+                return new Response(JSON.stringify({ result: { payload: asset } }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            if (url.includes('/points/delete')) {
+                return new Response(JSON.stringify({ status: 'ok' }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            throw new Error(`unexpected fetch: ${url}`);
+        });
+        globalThis.fetch = fetchMock;
+        const client = new MemoryClient();
+        const result = await client.deleteAsset(asset.id);
+        expect(result.deleted).toBe(true);
+        expect(result.asset?.name).toBe('ObsoletePattern');
+        expect(fetchMock.mock.calls.some(([url, init]) => String(url).includes('/points/delete') && init?.method === 'POST')).toBe(true);
+    });
+    test('deleteAsset 对不存在的资产返回 deleted=false', async () => {
+        vi.stubEnv('MEMORY_QDRANT_URL', 'http://127.0.0.1:50008');
+        const fetchMock = vi.fn(async (input) => {
+            const url = String(input);
+            if (url.includes('/points/missing-id')) {
+                return new Response(JSON.stringify({ result: null }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            throw new Error(`unexpected fetch: ${url}`);
+        });
+        globalThis.fetch = fetchMock;
+        const client = new MemoryClient();
+        const result = await client.deleteAsset('missing-id');
+        expect(result.deleted).toBe(false);
+        expect(result.asset).toBeNull();
+    });
+    test('updateAsset 更新已存在资产并保留 createdAt', async () => {
+        vi.stubEnv('MEMORY_QDRANT_URL', 'http://127.0.0.1:50008');
+        vi.stubEnv('MEMORY_EMBEDDING_URL', 'http://127.0.0.1:11434/api/embeddings');
+        vi.stubEnv('MEMORY_EMBEDDING_MODEL', 'nomic-embed-text');
+        const asset = {
+            id: 'asset-to-update',
+            name: 'PatternA',
+            type: 'pattern',
+            description: '旧描述',
+            summary: '旧摘要',
+            content: 'old snippet',
+            tags: ['pattern'],
+            confidence: 0.7,
+            createdAt: '2026-01-01T00:00:00.000Z',
+            updatedAt: '2026-01-01T00:00:00.000Z',
+        };
+        const fetchMock = vi.fn(async (input, init) => {
+            const url = String(input);
+            if (url.endsWith('/api/embeddings')) {
+                return new Response(JSON.stringify({ embedding: [0.1, 0.2, 0.3] }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            if (url.includes(`/points/${asset.id}`) && init?.method !== 'POST' && init?.method !== 'PUT') {
+                return new Response(JSON.stringify({ result: { payload: asset } }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            if (url.endsWith('/collections/mcp_probe_memory') && !init?.method) {
+                return new Response('', { status: 200 });
+            }
+            if (url.includes('/points?wait=true') && init?.method === 'PUT') {
+                return new Response(JSON.stringify({ status: 'ok' }), {
+                    status: 200,
+                    headers: { 'Content-Type': 'application/json' },
+                });
+            }
+            throw new Error(`unexpected fetch: ${url}`);
+        });
+        globalThis.fetch = fetchMock;
+        const client = new MemoryClient();
+        const result = await client.updateAsset(asset.id, { summary: '新摘要' });
+        expect(result.updated).toBe(true);
+        expect(result.asset?.summary).toBe('新摘要');
+        expect(result.asset?.createdAt).toBe(asset.createdAt);
+        expect(result.asset?.id).toBe(asset.id);
+        expect(fetchMock.mock.calls.some(([url, init]) => String(url).includes('/points?wait=true') && init?.method === 'PUT')).toBe(true);
+    });
 });

package/build/lib/__tests__/memory-orchestration.unit.test.d.ts

@@ -0,0 +1 @@
+export {};

package/build/lib/__tests__/memory-orchestration.unit.test.js

@@ -0,0 +1,88 @@
+import { afterEach, describe, expect, test, vi } from 'vitest';
+import { renderMemoryGuideSection, shouldShowSourceInSearch, } from '../memory-orchestration.js';
+afterEach(() => {
+    vi.unstubAllEnvs();
+});
+describe('memory-orchestration', () => {
+    test('does not show source path by default', () => {
+        vi.stubEnv('MEMORY_SEARCH_SHOW_SOURCE', '');
+        vi.stubEnv('MEMORY_REPO_ID', '');
+        const section = renderMemoryGuideSection({
+            enabled: true,
+            available: true,
+            degraded: false,
+            query: '404 submit',
+            results: [
+                {
+                    id: '1',
+                    score: 0.91,
+                    name: 'purchase-create-submit-404',
+                    type: 'bugfix',
+                    description: 'desc',
+                    summary: 'summary',
+                    content: '',
+                    tags: ['bugfix'],
+                    sourceProject: 'zhixing/gongyingshang',
+                    sourcePath: 'admin-api/app.js',
+                },
+            ],
+            assetsById: {},
+        });
+        expect(section).toContain('purchase-create-submit-404');
+        expect(section).not.toContain('admin-api/app.js');
+        expect(section).not.toContain('来源:');
+    });
+    test('shows source path when MEMORY_REPO_ID matches', () => {
+        vi.stubEnv('MEMORY_SEARCH_SHOW_SOURCE', '');
+        vi.stubEnv('MEMORY_REPO_ID', 'zhixing/gongyingshang');
+        expect(shouldShowSourceInSearch({
+            id: '1',
+            score: 0.9,
+            name: 'x',
+            type: 'bugfix',
+            description: '',
+            summary: '',
+            content: '',
+            tags: [],
+            sourceProject: 'zhixing/gongyingshang',
+            sourcePath: 'admin-api/app.js',
+        })).toBe(true);
+        expect(shouldShowSourceInSearch({
+            id: '2',
+            score: 0.9,
+            name: 'y',
+            type: 'bugfix',
+            description: '',
+            summary: '',
+            content: '',
+            tags: [],
+            sourceProject: 'other/repo',
+            sourcePath: 'src/index.ts',
+        })).toBe(false);
+    });
+    test('MEMORY_SEARCH_SHOW_SOURCE=true always shows source', () => {
+        vi.stubEnv('MEMORY_SEARCH_SHOW_SOURCE', 'true');
+        vi.stubEnv('MEMORY_REPO_ID', '');
+        const section = renderMemoryGuideSection({
+            enabled: true,
+            available: true,
+            degraded: false,
+            query: 'q',
+            results: [
+                {
+                    id: '1',
+                    score: 0.8,
+                    name: 'n',
+                    type: 'bugfix',
+                    description: '',
+                    summary: 's',
+                    content: '',
+                    tags: [],
+                    sourcePath: 'admin-api/app.js',
+                },
+            ],
+            assetsById: {},
+        });
+        expect(section).toContain('来源: admin-api/app.js');
+    });
+});

package/build/lib/__tests__/memory-payload.unit.test.d.ts

@@ -0,0 +1 @@
+export {};

package/build/lib/__tests__/memory-payload.unit.test.js

@@ -0,0 +1,35 @@
+import { describe, expect, test } from 'vitest';
+import { normalizeMemoryPayload, payloadToMemoryFields } from '../memory-payload.js';
+describe('memory-payload', () => {
+    test('maps legacy kind/title/source fields to current schema', () => {
+        const normalized = normalizeMemoryPayload({
+            kind: 'extracted_pattern',
+            title: 'Feishu submit success but sync_failed',
+            source: 'scan_and_extract_patterns-fallback',
+            content: 'Pattern: proxy mismatch',
+            tags: ['feishu', 'pattern'],
+            created_at: '2026-05-25T11:20:06.705Z',
+        });
+        expect(normalized.name).toBe('Feishu submit success but sync_failed');
+        expect(normalized.type).toBe('pattern');
+        expect(normalized.summary).toContain('Pattern');
+        expect(normalized.createdAt).toBe('2026-05-25T11:20:06.705Z');
+    });
+    test('payloadToMemoryFields preserves standard asset fields', () => {
+        const fields = payloadToMemoryFields({
+            id: 'asset-1',
+            name: 'purchase-create-submit-404',
+            type: 'bugfix',
+            description: '送审 404',
+            summary: 'res.data.purchase.id',
+            content: '【现象】404',
+            tags: ['bugfix'],
+            confidence: 0.95,
+            createdAt: '2026-05-27T04:28:04.684Z',
+            updatedAt: '2026-05-27T04:28:04.684Z',
+        });
+        expect(fields.name).toBe('purchase-create-submit-404');
+        expect(fields.type).toBe('bugfix');
+        expect(fields.tags).toEqual(['bugfix']);
+    });
+});

package/build/lib/__tests__/quality-constraints.unit.test.d.ts

@@ -0,0 +1 @@
+export {};