qwen-code/docs/design/auto-memory/memory-system.md
顾盼 9e2f63a1ca
feat(memory): managed auto-memory and auto-dream system (#3087)
* docs: add auto-memory implementation log

* feat(core): add managed auto-memory storage scaffold

* feat(core): load managed auto-memory index

* feat(core): add managed auto-memory recall

* feat(core): add managed auto-memory extraction

* feat(cli): add managed auto-memory dream commands

* feat(core): add auxiliary side-query foundation

* feat(memory): add model-driven recall selection

* feat(memory): add model-driven extraction planner

* feat(core): add background task runtime foundation

* feat(memory): schedule auto dream in background

* feat(core): add background agent runner foundation

* feat(memory): add extraction agent planner

* feat(core): add dream agent planner

* feat(core): rebuild managed memory index

* feat(memory): add governance status commands

* feat(memory): add managed forget flow

* feat(core): harden background agent planning

* feat(memory): complete managed parity closure

* test(memory): add managed lifecycle integration coverage

* feat: same to cc

* feat(memory-ui): add memory saved notification and memory count badge

Feature 3 - Memory Saved Notification:
- Add HistoryItemMemorySaved type to types.ts
- Create MemorySavedMessage component for rendering '● Saved/Updated N memories'
- In useGeminiStream: detect in-turn memory writes via mapToDisplay's
  memoryWriteCount field and emit 'memory_saved' history item after turn
- In client.ts: capture background dream/extract promises and expose
  via consumePendingMemoryTaskPromises(); useGeminiStream listens
  post-turn and emits 'Updated N memories' notification for background tasks

Feature 4 - Memory Count Badge:
- Add isMemoryOp field to IndividualToolCallDisplay
- Add memoryWriteCount/memoryReadCount to HistoryItemToolGroup
- Add detectMemoryOp() in useReactToolScheduler using isAutoMemPath
- ToolGroupMessage renders '● Recalled N memories, Wrote N memories' badge
  at the top of tool groups that touch memory files

Fix: process.env bracket-access in paths.ts (noPropertyAccessFromIndexSignature)
Fix: MemoryDialog.test.tsx mock useSettings to satisfy SettingsProvider requirement

* fix(memory-ui): auto-approve memory writes, collapse memory tool groups, fix MEMORY.md path

Problem 1 - Auto-approve memory file operations:
- write-file.ts: getDefaultPermission() checks isAutoMemPath; returns 'allow'
  for managed auto-memory files, 'ask' for all other files
- edit.ts: same pattern

Problem 2 - Feature 4 UX: collapse memory-only tool groups:
- ToolGroupMessage: detect when all tool calls have isMemoryOp set (pure memory
  group) and all are complete; render compact '● Recalled/Wrote N memories
  (ctrl+o to expand)' instead of individual tool call rows
- ctrl+o toggles expand/collapse when isFocused and group is memory-only
- Mixed groups (memory + other tools) keep badge-at-top behaviour
- Expanded state shows individual tool calls with '● Memory operations
  (ctrl+o to collapse)' header

Problem 3 - MEMORY.md path mismatch:
- prompt.ts: Step 2 now references full absolute path ${memoryDir}/MEMORY.md
  so the model writes to the correct location inside the memory directory,
  not to the parent project directory

Fix tests:
- write-file.test.ts: add getProjectRoot to mockConfigInternal
- prompt.test.ts: update assertion to match full-path section header

* fix(memory-ui): fix duplicate notification, broken ctrl+o, and Edit tool detection

- Remove duplicate 'Saved N memories' notification: the tool group badge already
  shows 'Wrote N memories'; the separate HistoryItemMemorySaved addItem after
  onComplete was double-counting. Keep only the background-task path
  (consumePendingMemoryTaskPromises).

- Remove ctrl+o expand: Ink's Static area freezes items on first render and
  cannot respond to user input. useInput/useState(isExpanded) in a Static item
  is a no-op. Removed the dead code; memory-only groups now always render as
  the compact summary (no fake interactive hint).

- Fix Edit tool detection: detectMemoryOp was checking for 'edit_file' but the
  real tool name constant is 'edit'. Also removed non-existent 'create_file'
  (write_file covers all writes). Now editing MEMORY.md is correctly identified
  as a memory write op, collapses to 'Wrote N memories', and is auto-approved.

* fix(dream): run /dream as a visible submit_prompt turn, not a silent background agent

The previous implementation ran an AgentHeadless background agent that could
take 5+ minutes with zero UI feedback — user saw a blank screen for the entire
duration and then at most one line of text.

Fix: /dream now returns submit_prompt with the consolidation task prompt so it
runs as a regular AI conversation turn. Tool calls (read_file, write_file, edit,
grep_search, list_directory, glob) are immediately visible as collapsed tool
groups as the model works through the memory files — identical UX to Claude Code.

Also export buildConsolidationTaskPrompt from dreamAgentPlanner so dreamCommand
can reuse the same detailed consolidation prompt that was already written.

* fix(memory): auto-allow ls/glob/grep on memory base directory

Add getMemoryBaseDir() to getDefaultPermission() allow list in ls.ts,
glob.ts, and grep.ts — mirrors the existing pattern in read-file.ts.

Without this, ListFiles/Glob/Grep on ~/.qwen/* would trigger an
approval dialog, blocking /dream at its very first step.

* fix(background): prevent permission prompt hangs in background agents

Match Claude Code's headless-agent intent: background memory agents must never
block on interactive permission prompts.

Wrap background runtime config so getApprovalMode() returns YOLO, ensuring any
ask decision is auto-approved instead of hanging forever. Add regression test
covering the wrapped approval mode.

* fix(memory): run auto extract through forked agent

Make managed auto-memory extraction follow the Claude Code architecture:
background extraction now uses a forked agent to read/write memory files
directly, instead of planning patches and applying them with a separate
filesystem pipeline.

Keep the old patch/model path only as fallback if the forked agent fails.
Add regression tests covering the new execution path and tool whitelist.

* refactor(memory): remove legacy extract fallback pipeline

Delete the old patch/model/heuristic extraction path entirely.
Managed auto-memory extract now runs only through the forked-agent
execution flow, with no planner/apply fallback stages remaining.

Also remove obsolete exports/tests and update scheduler/integration
coverage to use the forked-agent-only architecture.

* refactor(memory): move auxiliary files out of memory/ directory

meta.json, extract-cursor.json, and consolidation.lock are internal
bookkeeping files, not user-visible memories. Move them one level up
to the project state dir (parent of memory/) so that the memory/
directory contains only MEMORY.md and topic files, matching the
clean layout of the upstream reference implementation.

Add getAutoMemoryProjectStateDir() helper in paths.ts and update the
three path accessors + store.test.ts path assertions accordingly.

* fix(memory): record lastDreamAt after manual /dream run

The /dream command submits a prompt to the main agent (submit_prompt),
which writes memory files directly. Because it bypasses dreamScheduler,
meta.json was never updated and /memory always showed 'never'.

Fix by:
- Exporting writeDreamManualRunToMetadata() from dream.ts
- Adding optional onComplete callback to SubmitPromptActionReturn and
  SubmitPromptResult (types.ts / commands/types.ts)
- Propagating onComplete through slashCommandProcessor.ts
- Firing onComplete after turn completion in useGeminiStream.ts
- Providing the callback in dreamCommand.ts to write lastDreamAt

* fix(memory): remove scope params from /remember in managed auto-memory mode

--global/--project are legacy save_memory tool concepts. In managed
auto-memory mode the forked agent decides the appropriate type
(user/feedback/project/reference) based on the content of the fact.

Also improve the prompt wording to explicitly ask the agent to choose
the correct type, reducing the tendency to default to 'project'.

* feat(ui): show '✦ dreaming' indicator in footer during background dream

Subscribe to getManagedAutoMemoryDreamTaskRegistry() in Footer via a
useDreamRunning() hook. While any dream task for the current project is
pending or running, display '✦ dreaming' in the right section of the
footer bar, between Debug Mode and context usage.

* refactor(memory): align dream/extract infrastructure with Claude Code patterns

Five improvements based on Claude Code parity audit:

1. Memoize getAutoMemoryRoot (paths.ts)
   - Add _autoMemoryRootCache Map, keyed by projectRoot
   - findCanonicalGitRoot() walks the filesystem per call; memoize avoids
     repeated git-tree traversal on hot-path schedulers/scanners
   - Expose clearAutoMemoryRootCache() for test teardown

2. Lock file stores PID + isProcessRunning reclaim (dreamScheduler.ts)
   - acquireDreamLock() writes process.pid to the lock file body
   - lockExists() reads PID and calls process.kill(pid, 0); dead/missing
     PID reclaims the lock immediately instead of waiting 2h
   - Stale threshold reduced to 1h (PID-reuse guard, same as CC)

3. Session scan throttle (dreamScheduler.ts)
   - Add SESSION_SCAN_INTERVAL_MS = 10min (same as CC)
   - Add lastSessionScanAt Map<projectRoot, number> to ManagedAutoMemoryDreamRuntime
   - When time-gate passes but session-gate doesn't, throttle prevents
     re-scanning the filesystem on every user turn

4. mtime-based session counting (dreamScheduler.ts)
   - Replace fragile recentSessionIdsSinceDream Set in meta.json with
     filesystem mtime scan (listSessionsTouchedSince)
   - Mirrors Claude Code's listSessionsTouchedSince: reads session JSONL
     files from Storage.getProjectDir()/chats/, filters by mtime > lastDreamAt
   - Immune to meta.json corruption/loss; no per-turn metadata write
   - ManagedAutoMemoryDreamRuntime accepts injectable SessionScannerFn
     for clean unit testing without real session files

5. Extraction mutual exclusion extended to write_file/edit (extractScheduler.ts)
   - historySliceUsesMemoryTool() now checks write_file/edit/replace/create_file
     tool calls whose file_path is within isAutoMemPath()
   - Previously only detected save_memory; missed direct file writes by
     the main agent, causing redundant background extraction

* docs(memory): add user-facing memory docs, i18n for all locales, simplify /forget

- Add docs/users/features/memory.md: comprehensive user-facing guide covering
  QWEN.md instructions, auto-memory behaviour, all memory commands, and
  troubleshooting; replaces the placeholder auto-memory.md
- Update docs/users/features/_meta.ts: rename entry auto-memory → memory
- Update docs/users/features/commands.md: add /init, /remember, /forget,
  /dream rows; fix /memory description; remove /init duplicate
- Update docs/users/configuration/settings.md: add memory.* settings section
  (enableManagedAutoMemory, enableManagedAutoDream) between tools and permissions
- Remove /forget --apply flag: preview-then-apply flow replaced with direct
  deletion; update forgetCommand.ts, en.js, zh.js accordingly
- Add all auto-memory i18n keys to de, ja, pt, ru locales (18 keys each):
  Open auto-memory folder, Auto-memory/Auto-dream status lines, never/on/off,
  ✦ dreaming, /forget and /remember usage strings, all managed-memory messages
- Remove dead save_memory branch from extractScheduler.partWritesToMemory()
- Add ✦ dreaming indicator to Footer.tsx with i18n; fix Footer.test.tsx mocks
- Refactor MemoryDialog.tsx auto-dream status line to use i18n
- Remove save_memory tool (memoryTool.ts/test); clean up webui references
- Add extractionPlanner.ts, const.ts and associated tests
- Delete stale docs/users/configuration/memory.md and
  docs/developers/tools/memory.md (content superseded)

* refactor(memory): remove all Claude Code references from comments and test names

* test(memory): remove empty placeholder test files that cause vitest to fail

* fix eslint

* fix test in windows

* fix test

* fix(memory): address critical review findings from PR #3087

- fix(read-file): narrow auto-allow from getMemoryBaseDir() (~/.qwen) to
  isAutoMemPath(projectRoot) to prevent exposing settings.json / OAuth
  credentials without user approval (wenshao review)

- fix(forget): per-entry deletion instead of whole-file unlink
  - assign stable per-entry IDs (relativePath:index for multi-entry files)
    so the model can target individual entries without removing siblings
  - rewrite file keeping unmatched entries; only unlink when file becomes
    empty (wenshao review)

- fix(entries): round-trip correctness for multi-entry new-format bodies
  - parseAutoMemoryEntries: plain-text line closes current entry and opens
    a new one (was silently ignored when current was already set)
  - renderAutoMemoryBody: emit blank line between adjacent entries so the
    parser can detect entry boundaries on re-read (wenshao review)

- fix(entries): resolve two CodeQL polynomial-regex alerts
  - indentedMatch: \s{2,}(?:[-*]\s+)? → [\t ]{2,}(?:[-*][\t ]+)?
  - topLevelMatch: :\s*(.+)$ → :[ \t]*(\S.*)$
  (github-advanced-security review)

- fix(scan.test): use forward-slash literal for relativePath expectation
  since listMarkdownFiles() normalises all separators to '/' on all
  platforms including Windows

* fix(memory): replace isAutoMemPath startsWith with path.relative()

Using path.relative() instead of string startsWith() is more robust
across platforms — it correctly handles Windows path-separator
differences and avoids potential edge cases where a path prefix match
could succeed on non-separator boundaries.

Addresses github-actions review item 3 (PR #3087).

* feat(telemetry): add auto-memory telemetry instrumentation

Add OpenTelemetry logs + metrics for the five auto-memory lifecycle
events: extract, dream, recall, forget, and remember.

Telemetry layer (packages/core/src/telemetry/):
- constants.ts: 5 new event-name constants
  (qwen-code.memory.{extract,dream,recall,forget,remember})
- types.ts: 5 new event classes with typed constructor params
  (MemoryExtractEvent, MemoryDreamEvent, MemoryRecallEvent,
   MemoryForgetEvent, MemoryRememberEvent)
- metrics.ts: 8 new OTel instruments (5 Counters + 3 Histograms)
  with recordMemoryXxx() helpers; registered inside initializeMetrics()
- loggers.ts: logMemoryExtract/Dream/Recall/Forget/Remember() — each
  emits a structured log record and calls its recordXxx() counterpart
- index.ts: re-exports all new symbols

Instrumentation call-sites:
- extractScheduler.ts ManagedAutoMemoryExtractRuntime.runTask():
  emits extract event with trigger=auto, completed/failed status,
  patches_count, touched_topics, and wall-clock duration
- dream.ts runManagedAutoMemoryDream():
  emits dream event with trigger=auto, updated/noop status,
  deduped_entries, touched_topics, and duration; covers both
  agent-planner and mechanical fallback paths
- recall.ts resolveRelevantAutoMemoryPromptForQuery():
  emits recall event with strategy, docs_scanned/selected, and
  duration; covers model, heuristic, and none paths
- forget.ts forgetManagedAutoMemoryEntries():
  emits forget event with removed_entries_count, touched_topics,
  and selection_strategy (model/heuristic/none)
- rememberCommand.ts action():
  emits remember event with topic=managed|legacy at command
  invocation time (before agent decides the actual memory type)

* refactor(telemetry): remove memory forget/remember telemetry events

Remove EVENT_MEMORY_FORGET and EVENT_MEMORY_REMEMBER along with all
associated infrastructure that is no longer needed:

- constants.ts: remove EVENT_MEMORY_FORGET, EVENT_MEMORY_REMEMBER
- types.ts: remove MemoryForgetEvent, MemoryRememberEvent classes
- metrics.ts: remove MEMORY_FORGET_COUNT, MEMORY_REMEMBER_COUNT constants,
  memoryForgetCounter, memoryRememberCounter module vars,
  their initialization in initializeMetrics(), and
  recordMemoryForgetMetrics(), recordMemoryRememberMetrics() functions
- loggers.ts: remove logMemoryForget(), logMemoryRemember() functions
  and their imports
- index.ts: remove all re-exports for the above symbols
- memory/forget.ts: remove logMemoryForget call-site and import
- cli/rememberCommand.ts: remove logMemoryRemember call-sites and import

* change default value

* fix forked agent

* refactor(background): unify fork primitives into runForkedAgent + cleanup

- Merge runForkedQuery into runForkedAgent via TypeScript overloads:
  with cacheSafeParams → GeminiChat single-turn path (ForkedQueryResult)
  without cacheSafeParams → AgentHeadless multi-turn path (ForkedAgentResult)
- Delete forkedQuery.ts; move its test to background/forkedAgent.cache.test.ts
- Remove forkedQuery export from followup/index.ts
- Migrate all callers (suggestionGenerator, speculation, btwCommand, client)
  to import from background/forkedAgent
- Add getFastModel() / setFastModel() to Config; expose in CLI config init
  and ModelDialog / modelCommand
- Remove resolveFastModel() from AppContainer — now delegated to config.getFastModel()
- Strip Claude Code references from code comments

* fix(memory): address wenshao's critical review findings

- dream.ts: writeDreamManualRunToMetadata now persists lastDreamSessionId
  and resets recentSessionIdsSinceDream, preventing auto-dream from firing
  again in the same session after a manual /dream
- config.ts: gate managed auto-memory injection on getManagedAutoMemoryEnabled();
  when disabled, previously saved memories are no longer injected into new sessions
- rememberCommand.ts: remove legacy save_memory branch (tool was removed);
  fall back to submit_prompt directing agent to write to QWEN.md instead
- BuiltinCommandLoader.ts: only register /dream and /forget when managed
  auto-memory is enabled, matching the feature's runtime availability
- forget.ts: return early in forgetManagedAutoMemoryMatches when matches is
  empty, avoiding unnecessary directory scaffolding as a side effect

* fix test

* fix ci test

* feat(memory): align extract/dream agents to Claude Code patterns

- fix(client): move saveCacheSafeParams before early-return paths so
  extract agents always have cache params available (fixes extract never
  triggering in skipNextSpeakerCheck mode)

- feat(extract): add read-only shell tool + memory-scoped write
  permissions; create inline createMemoryScopedAgentConfig() with
  PermissionManager wrapper (isToolEnabled + evaluate) that allows only
  read-only shell commands and write/edit within the auto-memory dir

- feat(extract): align prompt to Claude Code patterns — manifest block
  listing existing files, parallel read-then-write strategy, two-step
  save (memory file then index)

- feat(dream): remove mechanical fallback; runManagedAutoMemoryDream is
  now agent-only and throws without config

- feat(dream): align prompt to Claude Code 4-phase structure
  (Orient/Gather/Consolidate/Prune+Index); add narrow transcript grep,
  relative→absolute date conversion, stale index pruning, index size cap

- fix(permissions): add isToolEnabled() to MemoryScopedPermissionManager
  to prevent TypeError crash in CoreToolScheduler._schedule

- test: update dreamScheduler tests to mock dream.js; replace removed
  mechanical-dedup test with scheduler infrastructure verification

* move doc to design

* refactor(memory): unify extract+dream background task management into MemoryBackgroundTaskHub

- Add memoryTaskHub.ts: single BackgroundTaskRegistry + BackgroundTaskDrainer shared
  by all memory background tasks; exposes listExtractTasks() / listDreamTasks()
  typed query helpers and a unified drain() method
- extractScheduler: ManagedAutoMemoryExtractRuntime accepts hub via constructor
  (defaults to defaultMemoryTaskHub); test factory gets isolated fresh hub
- dreamScheduler: same pattern — sessionScanner + hub injection; BackgroundTask-
  Scheduler initialized from injected hub; test factory gets isolated hub
- status.ts: replace two separate getRegistry() calls with defaultMemoryTaskHub
  typed query methods
- Footer.tsx (useDreamRunning): subscribe to shared registry, filter by
  DREAM_TASK_TYPE so extract tasks do not trigger the dream spinner
- index.ts: re-export memoryTaskHub.ts so defaultMemoryTaskHub/DREAM_TASK_TYPE/
  EXTRACT_TASK_TYPE are available as top-level package exports

* refactor(background): introduce general-purpose BackgroundTaskHub

Replace memory-specific MemoryBackgroundTaskHub with a domain-agnostic
BackgroundTaskHub in the background/ layer. Any future background task
runtime (3rd, 4th, …) plugs in by accepting a hub via constructor
injection — no new infrastructure required.

Changes:
- Add background/taskHub.ts: BackgroundTaskHub (registry + drainer +
  createScheduler() + listByType(taskType, projectRoot?)) and the
  globalBackgroundTaskHub singleton. Zero knowledge of any task type.
- Delete memory/memoryTaskHub.ts: its narrow listExtractTasks /
  listDreamTasks helpers are replaced by the generic listByType() call.
- Move EXTRACT_TASK_TYPE to extractScheduler.ts (owned by the runtime
  that defines it); replace 3 hardcoded string literals with the const.
- Move DREAM_TASK_TYPE to dreamScheduler.ts; use hub.createScheduler()
  instead of manually wiring new BackgroundTaskScheduler(reg, drain).
- status.ts: globalBackgroundTaskHub.listByType(EXTRACT_TASK_TYPE, ...)
- Footer.tsx: globalBackgroundTaskHub.registry (shared, filtered by type)
- index.ts: export background/taskHub.js; drop memory/memoryTaskHub.js

* test(background): add BackgroundTaskHub unit tests and hub isolation checks

- background/taskHub.test.ts (11 tests):
  - createScheduler(): tasks registered via scheduler appear in hub registry;
    multiple calls return distinct scheduler instances
  - listByType(): filters by taskType, filters by projectRoot, returns []
    for unknown types, two types co-exist in registry but stay separated
  - drain(): resolves false on timeout, resolves true when tasks complete,
    resolves true immediately when no tasks in flight
  - isolation: tasks in hubA do not appear in hubB
  - globalBackgroundTaskHub: is a BackgroundTaskHub instance with registry/drainer

- extractScheduler.test.ts (+1 test):
  - factory-created runtimes have isolated registries; tasks in runtimeA
    are invisible to runtimeB; all tasks carry EXTRACT_TASK_TYPE

- dreamScheduler.test.ts (+1 test):
  - factory-created runtimes have isolated registries; tasks in runtimeA
    are invisible to runtimeB; all tasks carry DREAM_TASK_TYPE

* refactor(memory): consolidate all memory state into MemoryManager

Replace BackgroundTaskRegistry/Drainer/Scheduler/Hub helper classes and
module-level globals with a single MemoryManager class owned by Config.

## Changes

### New
- packages/core/src/memory/manager.ts — MemoryManager with:
  - scheduleExtract / scheduleDream (inline queuing + deduplication logic)
  - recall / forget / selectForgetCandidates / forgetMatches
  - getStatus / drain / appendToUserMemory
  - subscribe(listener) compatible with useSyncExternalStore
  - storeWith() atomic record registration (no double-notify)
  - Distinct skippedReason 'scan_throttled' vs 'min_sessions' for dream
- packages/core/src/utils/forkedAgent.ts — pure cache util (moved from background/)
- packages/core/src/utils/sideQuery.ts — pure util (moved from auxiliary/)

### Deleted
- background/taskRegistry, taskDrainer, taskScheduler, taskHub and all tests
- background/forkedAgent (moved to utils/)
- auxiliary/sideQuery (moved to utils/)
- memory/extractScheduler, dreamScheduler, state and all tests

### Modified
- config/config.ts — Config owns MemoryManager instance; getMemoryManager()
- core/client.ts — all memory ops via config.getMemoryManager()
- core/client.test.ts — mock MemoryManager instead of individual modules
- memory/status.ts — accepts MemoryManager param, drops globalBackgroundTaskHub
- index.ts — memory exports reduced from 14 modules to 5 (manager/types/paths/store/const)
- cli/commands/dreamCommand.ts — via config.getMemoryManager()
- cli/commands/forgetCommand.ts — via config.getMemoryManager()
- cli/components/Footer.tsx — useSyncExternalStore replacing setInterval polling
- cli/components/Footer.test.tsx — add getMemoryManager mock
2026-04-16 20:05:45 +08:00

23 KiB
Raw Blame History

Memory 记忆管理系统

本文介绍 Qwen Code 中 Managed Auto-Memory(托管自动记忆)的记忆管理机制、触发时机和实现细节。


目录

  1. 概述
  2. 存储结构
  3. 记忆类型
  4. 记忆条目格式
  5. 核心生命周期
  6. Extract — 提取
  7. Dream — 整合
  8. Recall — 召回
  9. Forget — 遗忘
  10. 索引重建
  11. 遥测埋点

概述

Managed Auto-Memory 是一套在 AI 会话过程中自动积累、整合和检索用户相关知识的持久化记忆系统。它通过四个核心操作维护记忆的生命周期:

操作 英文 触发方式 作用
提取 Extract 自动(每轮对话后) 从对话记录中提炼新知识写入记忆文件
整合 Dream 自动(周期性后台任务) 对记忆文件去重、合并,保持整洁
召回 Recall 自动(每轮对话前) 检索与当前请求相关的记忆注入到系统提示
遗忘 Forget 手动(用户命令 /forget 精确删除指定的记忆条目

存储结构

目录布局

~/.qwen/                                      ← 全局基础目录(默认)
└── projects/
    └── <sanitized-git-root>/                 ← 项目标识(基于 Git 根路径)
        ├── meta.json                         ← 元数据(提取/整合时间戳、状态)
        ├── extract-cursor.json               ← 提取游标(已处理的对话偏移量)
        ├── consolidation.lock                ← Dream 进程互斥锁
        └── memory/                           ← 记忆主目录
            ├── MEMORY.md                     ← 索引文件(自动生成,汇总所有条目)
            ├── user.md                       ← 用户偏好记忆(示例)
            ├── feedback.md                   ← 反馈规范记忆(示例)
            ├── project/
            │   └── milestone.md              ← 项目记忆(支持子目录)
            └── reference/
                └── grafana.md                ← 外部资源记忆

环境变量覆盖

  • QWEN_CODE_MEMORY_BASE_DIR:替换全局基础目录
  • QWEN_CODE_MEMORY_LOCAL=1:改用项目内路径 .qwen/memory/

关键文件说明

文件 说明
meta.json 记录最后一次 Extract / Dream 的时间、会话 ID、涉及的记忆类型、执行状态
extract-cursor.json 记录当前会话已处理到对话历史的哪个偏移量,避免重复提取
consolidation.lock Dream 运行时的文件锁,内容为持有者 PID超过 1 小时自动失效
MEMORY.md 所有主题文件的索引,每次 Extract/Dream 后重建,格式为 Markdown 列表

记忆类型

系统支持四种内置记忆类型,每种对应不同的信息维度:

类型 存储内容 何时写入 何时读取
user 用户的角色、技能背景、工作习惯 了解到用户角色/偏好/知识背景时 回答需要根据用户背景定制时
feedback 用户对 AI 行为的指导:避免什么、继续什么 用户纠正 AI 或确认某种非显而易见的做法时 影响 AI 行为方式时
project 项目进展、目标、决策、截止日期、Bug 追踪 了解到谁在做什么、为什么、截止何时时 帮助 AI 理解工作背景和动机时
reference 外部系统资源指针Dashboard、工单系统、Slack 频道等) 得知某种外部资源及其用途时 用户提及外部系统或相关信息时

不应该存入记忆的内容:代码模式/约定、Git 历史、调试方案、临时任务状态、已在 QWEN.md/AGENTS.md 中记录的内容。


记忆条目格式

每个主题文件使用 YAML frontmatter + Markdown body 格式:

---
name: 记忆名称
description: 一句话描述(用于判断召回相关性,要具体)
type: user|feedback|project|reference
---

记忆主体内容summary 行)

Why: 背后原因(让 AI 能理解边界情况而不是盲目遵守规则)
How to apply: 适用场景和使用方式

对于 feedbackproject 类型,强烈建议填写 WhyHow to apply,使记忆在边界情况下仍能正确应用。


核心生命周期

flowchart TD
    A([用户发送请求]) --> B

    subgraph "召回 Recall"
        B[扫描所有主题文件] --> C{文档数量和\n查询内容是否有效?}
        C -- 否 --> D[返回空提示词\nstrategy: none]
        C -- 是 --> E{是否配置了 Config?}
        E -- 是 --> F[模型驱动选择\nside query]
        F --> G{选出相关文档?}
        G -- 是 --> H[strategy: model]
        G -- 否 --> I[strategy: none]
        E -- 否 --> J[启发式关键词评分]
        F -- 失败 --> J
        J --> K{有得分 > 0 的文档?}
        K -- 是 --> L[strategy: heuristic]
        K -- 否 --> I
        H --> M[构建 Relevant Memory 提示词\n注入系统提示]
        L --> M
        I --> N[不注入记忆]
    end

    M --> O([AI 处理请求])
    N --> O
    D --> O

    O --> P([AI 返回响应])

    subgraph "提取 Extract后台"
        P --> Q{本轮 AI 是否\n直接写了记忆文件?}
        Q -- 是 --> R[跳过\nmemory_tool]
        Q -- 否 --> S{提取任务是否\n正在运行?}
        S -- 是 --> T[放入队列或跳过\nalready_running / queued]
        S -- 否 --> U[加载未处理的对话切片\n基于 extract cursor]
        U --> V[调用提取 Agent\nrunAutoMemoryExtractionByAgent]
        V --> W[去重规范化 patches]
        W --> X{有 touched topics?}
        X -- 是 --> Y[更新 meta.json\n重建 MEMORY.md 索引]
        X -- 否 --> Z[仅更新 extract cursor]
        Y --> Z
    end

    subgraph "Dream 整合(后台,周期性)"
        P --> AA{Dream 调度门控检查}
        AA --> AB{是否同一会话?}
        AB -- 是 --> AC[跳过\nsame_session]
        AB -- 否 --> AD{距上次 Dream\n≥ 24 小时?}
        AD -- 否 --> AE[跳过\nmin_hours]
        AD -- 是 --> AF{距上次 Dream 后\n新会话数 ≥ 5?}
        AF -- 否 --> AG[跳过\nmin_sessions]
        AF -- 是 --> AH{consolidation.lock\n是否存在?}
        AH -- 是 --> AI[跳过\nlocked]
        AH -- 否 --> AJ[获取锁\n写入 PID]
        AJ --> AK{是否配置了 Config?}
        AK -- 是 --> AL[Agent 路径\nplanManagedAutoMemoryDreamByAgent]
        AL --> AM{Agent 是否触碰了文件?}
        AM -- 是 --> AN[记录触碰的 topics]
        AM -- "否/失败" --> AO
        AK -- 否 --> AO[机械去重路径\n解析+去重+按字母排序]
        AO --> AP[写回更新后的主题文件]
        AN --> AQ[重建 MEMORY.md 索引\n更新 meta.json]
        AP --> AQ
        AQ --> AR[释放锁]
    end

Extract — 提取

触发时机

每次 AI 完成一轮响应后,由 scheduleAutoMemoryExtract 自动触发(后台非阻塞)。

调度逻辑(extractScheduler.ts

flowchart TD
    A[scheduleAutoMemoryExtract 被调用] --> B{本轮历史记录中\n是否有写记忆文件的工具调用?}
    B -- 是 --> C[登记 skipped 任务\n原因: memory_tool]
    B -- 否 --> D{isExtractRunning?}
    D -- 是 --> E{是否已有 queued 请求?}
    E -- 是 --> F[更新 queued 请求的\nhistory 参数]
    E -- 否 --> G[注册 pending 任务\n放入 queue]
    D -- 否 --> H[注册 running 任务\n调用 runTask]
    H --> I[markExtractRunning\nsetCurrentTaskId]
    I --> J[runAutoMemoryExtract]
    J --> K[任务完成]
    K --> L[clearExtractRunning\n检查 queue → startQueuedIfNeeded]
    F --> M[返回 skipped: queued]
    G --> M
    C --> N[返回 skipped: memory_tool]

跳过原因说明

原因 含义
memory_tool 本轮主 Agent 已直接写了记忆文件,跳过以避免冲突
already_running 提取正在进行且无法入队
queued 已有提取在运行,本次请求已入队

核心提取流程(extract.ts

flowchart TD
    A[runAutoMemoryExtract] --> B[ensureAutoMemoryScaffold\n初始化目录和文件]
    B --> C[buildTranscriptMessages\n将 Content[] 转换为带 offset 的消息列表]
    C --> D[readExtractCursor\n读取上次处理到的位置]
    D --> E[loadUnprocessedTranscriptSlice\n截取未处理的消息段]
    E --> F{slice 为空?}
    F -- 是 --> G[返回无 patches 结果]
    F -- 否 --> H[runAutoMemoryExtractionByAgent\n调用 forked agent 提取 patches]
    H --> I[dedupeExtractPatches\n去重+规范化]
    I --> J{有 touched topics?}
    J -- 是 --> K[bumpMetadata\n更新 meta.json]
    K --> L[rebuildManagedAutoMemoryIndex\n重建 MEMORY.md]
    L --> M[writeExtractCursor\n记录最新 offset]
    J -- 否 --> M
    M --> N[返回 AutoMemoryExtractResult]

提取游标Cursor

  • 字段:{ sessionId, processedOffset, updatedAt }
  • 每次提取后更新 processedOffset 为当前历史长度
  • 下次提取时,只处理 offset >= processedOffset 的消息
  • 跨会话时(sessionId 变化)从偏移量 0 重新开始

Patch 过滤规则

  • 摘要长度 < 12 字符 → 丢弃
  • 摘要以 ? 结尾 → 丢弃(疑问句)
  • 包含临时性关键词today/now/currently/temporary 等)→ 丢弃
  • 相同 topic:summary 组合 → 去重

Dream — 整合

触发时机

每次 AI 完成一轮响应后,由 scheduleManagedAutoMemoryDream 自动触发(后台非阻塞)。但受多个门控条件保护,大多数情况下会被跳过。

调度门控(dreamScheduler.ts

flowchart TD
    A[scheduleManagedAutoMemoryDream 被调用] --> B{Dream 功能是否启用?}
    B -- 否 --> C[跳过: disabled]
    B -- 是 --> D[ensureAutoMemoryScaffold\n读取 lastDreamSessionId]
    D --> E{当前 sessionId\n== lastDreamSessionId?}
    E -- 是 --> F[跳过: same_session]
    E -- 否 --> G{elapsedHours ≥ 24h\n或从未 dream?}
    G -- 否 --> H[跳过: min_hours]
    G -- 是 --> I{距上次 session scan\n< 10 分钟?}
    I -- 是 --> J[跳过: min_sessions\n等待下次扫描窗口]
    I -- 否 --> K[扫描 chats/*.jsonl mtime\n统计上次 Dream 后的新会话数]
    K --> L{新会话数 ≥ 5?}
    L -- 否 --> M[跳过: min_sessions]
    L -- 是 --> N{lockExists?\nPID 检查 + 过期检查}
    N -- 是 --> O[跳过: locked]
    N -- 否 --> P{dedupeKey 是否已有\n同项目 Dream 任务?}
    P -- 是 --> Q[跳过: running\n返回已有 taskId]
    P -- 否 --> R[调度后台任务\nBgTaskScheduler]
    R --> S[acquireDreamLock\n写入 PID 到 consolidation.lock]
    S --> T[runManagedAutoMemoryDream]
    T --> U[更新 meta.json\n释放锁]

门控参数

参数 默认值 说明
minHoursBetweenDreams 24 小时 两次 Dream 之间的最小时间间隔
minSessionsBetweenDreams 5 个会话 触发 Dream 所需的最小新会话数
SESSION_SCAN_INTERVAL_MS 10 分钟 会话文件扫描的节流间隔
DREAM_LOCK_STALE_MS 1 小时 lock 文件被视为过期的时间阈值

锁机制

  • lock 文件位于 <project-state-dir>/consolidation.lock
  • 内容为持有进程的 PID
  • 检查时:若 PID 进程已不存在(kill(pid, 0) 失败)或 lock 超过 1 小时 → 视为过期,自动清除

整合执行流程(dream.ts

flowchart TD
    A[runManagedAutoMemoryDream] --> B{是否配置了 Config?}
    B -- 是 --> C[Agent 路径\nplanManagedAutoMemoryDreamByAgent]
    C --> D{Agent 是否修改了文件?}
    D -- 是 --> E[从文件路径推断 touched topics]
    E --> F[bumpMetadata\n重建 MEMORY.md 索引]
    F --> G[updateDreamMetadataResult]
    G --> H[记录遥测事件]
    H --> I[返回结果]
    B -- 否 --> J[机械去重路径]
    C -- 抛出异常 --> J
    D -- 否 --> J

    J --> K[scanAutoMemoryTopicDocuments\n读取所有主题文件]
    K --> L[对每个文件执行 buildDreamedBody]
    L --> M[解析 entries → 按 summary 去重\n按字母升序排序 → 重新渲染]
    M --> N{body 有变化?}
    N -- 是 --> O[写回文件]
    O --> P[记录 touched topic]
    N --> Q[检查跨文件重复\ndedupeKey = type:summary]
    Q --> R{发现重复文件?}
    R -- 是 --> S[合并 entries 到 canonical 文件\n删除重复文件]
    S --> P
    R -- 否 --> T{有 touched topics?}
    P --> T
    T -- 是 --> U[bumpMetadata\n重建 MEMORY.md 索引]
    U --> V[updateDreamMetadataResult\n记录遥测 → 返回结果]
    T -- 否 --> V

机械去重逻辑

  1. 对每个主题文件内部:按 summary.toLowerCase() 去重,合并 why/howToApply 字段
  2. 按 summary 字母顺序重新排序
  3. 跨文件:相同 type:summary 的条目合并到最先发现的文件,删除重复文件

Recall — 召回

触发时机

每轮 AI 处理用户请求之前,由 resolveRelevantAutoMemoryPromptForQuery 自动触发,将相关记忆注入系统提示词。

召回流程(recall.ts

flowchart TD
    A[resolveRelevantAutoMemoryPromptForQuery] --> B[scanAutoMemoryTopicDocuments\n扫描所有主题文件]
    B --> C[filterExcludedAutoMemoryDocuments\n过滤本轮已写入的文件]
    C --> D{query 为空\n或 docs 为空\n或 limit <= 0?}
    D -- 是 --> E[返回空 prompt\nstrategy: none]
    D -- 否 --> F{是否配置了 Config?}
    F -- 是 --> G[selectRelevantAutoMemoryDocumentsByModel\n发起 side query 请求模型选择]
    G --> H{模型返回结果?}
    H -- 有文档 --> I[strategy: model]
    H -- 无文档 --> J[strategy: none\n仍然返回空]
    G -- "失败/异常" --> K[回退到启发式选择]
    F -- 否 --> K
    K --> L[tokenize query\n提取 ≥3 字符的 token]
    L --> M[scoreDocument 打分\n关键词匹配 +2 / 类型关键词 +1 / 有内容 +1]
    M --> N[过滤 score=0 的文档\n按分数降序排列取 Top 5]
    N --> O{有得分文档?}
    O -- 是 --> P[strategy: heuristic]
    O -- 否 --> J
    I --> Q[buildRelevantAutoMemoryPrompt\n构建 Relevant Memory 区块]
    P --> Q
    Q --> R[返回注入主系统提示的 prompt 片段]

评分规则(启发式)

条件 加分
query token 出现在文档内容中 +2每个 token
query token 是该类型的特征关键词 +1每个 token
文档 body 非空 +1

每种类型的特征关键词

  • useruser, preference, background, role, terse
  • feedbackfeedback, rule, avoid, style, summary
  • projectproject, goal, incident, deadline, release
  • referencereference, dashboard, ticket, docs, link

Prompt 构建规则

  • 最多注入 5 篇文档(MAX_RELEVANT_DOCS
  • 每篇文档 body 截断至 1200 字符(MAX_DOC_BODY_CHARS
  • 超出截断时追加提示:"NOTE: Relevant memory truncated for prompt budget."
  • 包含文档的新鲜度信息(基于文件 mtime

Forget — 遗忘

触发时机

由用户手动执行 /forget <query> 命令触发。

遗忘流程(forget.ts

flowchart TD
    A[forgetManagedAutoMemoryEntries\nquery + config] --> B[ensureAutoMemoryScaffold]
    B --> C[listIndexedForgetCandidates\n扫描所有文件的所有 entry]
    C --> D[为每个 entry 生成稳定 ID\n单 entry 文件: relativePath\n多 entry 文件: relativePath:index]
    D --> E{是否配置了 Config?}
    E -- 是 --> F[selectByModel\n构建 selection prompt\n发起 side query temperature=0]
    F --> G{模型选择成功?}
    G -- 是 --> H[strategy: model]
    G -- 失败 --> I[selectByHeuristic\n关键词匹配]
    E -- 否 --> I
    I --> J[strategy: heuristic]
    H --> K[遍历选中的 candidates]
    J --> K
    K --> L{entries.length == 1?}
    L -- 是 --> M[删除整个文件\nfs.unlink]
    L -- 否 --> N[解析文件中的所有 entries\n移除目标 entry\n重新渲染写回]
    M --> O[记录 removedEntries]
    N --> O
    O --> P{有 touched topics?}
    P -- 是 --> Q[bumpMetadata\n重建 MEMORY.md 索引]
    P --> R[返回 AutoMemoryForgetResult]
    Q --> R

Entry ID 设计

  • 单条目文件(常见情况):relativePath(如 feedback/no-summary.md
  • 多条目文件:relativePath:index(如 feedback/style.md:2
  • 使用稳定 ID 使模型可以精确定位条目而不影响同文件的其他条目

索引重建

MEMORY.md 是所有主题文件的导航索引,每次 Extract 或 Dream 后调用 rebuildManagedAutoMemoryIndex 重建:

- [用户偏好](user/preferences.md) — 用户是资深 Go 工程师,第一次接触 React
- [反馈规范](feedback/style.md) — 保持回复简洁,不要尾部总结
- [项目里程碑](project/milestone.md) — 移动端发布切分支前的合并冻结窗口

索引限制

  • 每行最多 150 字符(超出用 截断)
  • 最多 200 行
  • 总大小不超过 25,000 字节

遥测埋点

系统内置三类遥测事件,用于监控记忆操作的性能和效果:

Extract 遥测

字段 类型 说明
trigger 'auto' 触发方式(当前仅自动)
status 'completed' | 'failed' 执行结果
patches_count number 提取到的有效 patch 数量
touched_topics string[] 被写入的记忆类型列表
duration_ms number 总耗时(毫秒)

Dream 遥测

字段 类型 说明
trigger 'auto' 触发方式
status 'updated' | 'noop' | 'failed' 执行结果
deduped_entries number 机械路径去重的条目数量
touched_topics string[] 被修改的记忆类型列表
duration_ms number 总耗时(毫秒)

Recall 遥测

字段 类型 说明
query_length number 查询字符串长度
docs_scanned number 扫描的文档总数
docs_selected number 最终注入的文档数
strategy 'none' | 'heuristic' | 'model' 选择策略
duration_ms number 总耗时(毫秒)

相关源文件索引

文件 职责
packages/core/src/memory/types.ts 类型定义:AutoMemoryTypeAutoMemoryMetadataAutoMemoryExtractCursor
packages/core/src/memory/paths.ts 路径计算:getAutoMemoryRootisAutoMemPath、各类文件路径 helpers
packages/core/src/memory/store.ts 脚手架初始化:ensureAutoMemoryScaffold,索引/元数据读写
packages/core/src/memory/scan.ts 扫描主题文件:scanAutoMemoryTopicDocuments,解析 frontmatter
packages/core/src/memory/entries.ts 条目解析和渲染:parseAutoMemoryEntriesrenderAutoMemoryBody
packages/core/src/memory/extract.ts 提取核心逻辑:runAutoMemoryExtract游标管理patch 去重
packages/core/src/memory/extractScheduler.ts 提取调度器:ManagedAutoMemoryExtractRuntime,队列/运行状态机
packages/core/src/memory/extractionAgentPlanner.ts 提取 AgentrunAutoMemoryExtractionByAgent
packages/core/src/memory/dream.ts 整合核心逻辑:runManagedAutoMemoryDreamAgent 路径 + 机械去重
packages/core/src/memory/dreamScheduler.ts 整合调度器:ManagedAutoMemoryDreamRuntime,门控检查,锁管理
packages/core/src/memory/dreamAgentPlanner.ts 整合 AgentplanManagedAutoMemoryDreamByAgent
packages/core/src/memory/recall.ts 召回逻辑:resolveRelevantAutoMemoryPromptForQuery,启发式+模型双路径
packages/core/src/memory/forget.ts 遗忘逻辑:forgetManagedAutoMemoryEntries,候选生成+精确删除
packages/core/src/memory/indexer.ts 索引重建:rebuildManagedAutoMemoryIndexbuildManagedAutoMemoryIndex
packages/core/src/memory/prompt.ts 系统提示模板:记忆类型说明、格式示例、使用规范
packages/core/src/memory/governance.ts 治理建议类型:AutoMemoryGovernanceSuggestionType
packages/core/src/memory/state.ts 提取运行状态:isExtractRunningmarkExtractRunningclearExtractRunning
packages/core/src/memory/memoryAge.ts 新鲜度描述:memoryAgememoryFreshnessText