memall-os 0.1.2__tar.gz

memall_os-0.1.2/.env.example

@@ -0,0 +1,14 @@
+# ── MemALL 环境变量配置 ──
+# Copy this file to .env and fill in your values.
+
+# 必需
+# MEMALL_CHAT_ID=<飞书群聊chat_id>
+# TAVILY_API_KEY=<Tavily API密钥，用于网络搜索>
+# FEISHU_APP_ID=<飞书应用ID>
+# FEISHU_APP_SECRET=<飞书应用密钥>
+# OPENROUTER_API_KEY=<OpenRouter API密钥，用于LLM调用>
+
+# 可选
+# MEMALL_LOG_LEVEL=INFO
+# MEMALL_DB_PATH=data.db
+# MEMALL_BRIDGE_ENABLED=true

memall_os-0.1.2/.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,30 @@
+name: Bug Report
+description: Report a bug to help us improve
+labels: ["bug"]
+body:
+  - type: textarea
+    id: description
+    attributes:
+      label: Describe the bug
+      description: What happened and what did you expect?
+    validations:
+      required: true
+  - type: textarea
+    id: steps
+    attributes:
+      label: Steps to reproduce
+      description: How can we reproduce this?
+    validations:
+      required: true
+  - type: input
+    id: version
+    attributes:
+      label: MemALL version
+      description: Run `memall --version` or check `pip show memall`
+  - type: dropdown
+    id: os
+    attributes:
+      label: OS
+      options: [Windows, macOS, Linux]
+    validations:
+      required: true

memall_os-0.1.2/.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,23 @@
+name: Feature Request
+description: Suggest an idea for MemALL
+labels: ["enhancement"]
+body:
+  - type: textarea
+    id: problem
+    attributes:
+      label: Problem
+      description: What problem does this solve?
+    validations:
+      required: true
+  - type: textarea
+    id: solution
+    attributes:
+      label: Proposed solution
+      description: How should this work?
+    validations:
+      required: true
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: Alternatives
+      description: What else have you considered?

memall_os-0.1.2/.github/workflows/ci.yml

@@ -0,0 +1,36 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.10", "3.11"]
+
+    steps:
+      - uses: actions/checkout@v4
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e .
+          pip install pytest flake8 mypy
+      - name: Lint with flake8
+        run: |
+          flake8 src/memall/ --count --statistics --max-line-length=120
+      - name: Test with pytest
+        run: |
+          python -m pytest tests/ -v --tb=short
+      - name: Type check with mypy
+        continue-on-error: true
+        run: |
+          mypy src/memall/ --ignore-missing-imports --follow-imports=silent --python-version 3.10

memall_os-0.1.2/.gitignore

@@ -0,0 +1,155 @@
+# ── Python ──
+.venv/
+__pycache__/
+*.pyc
+*.pyo
+*.egg-info/
+.eggs/
+dist/
+build/
+*.so
+
+# ── IDE / Editor ──
+.idea/
+.vscode/
+*.swp
+*.swo
+*~
+
+# ── OS ──
+.DS_Store
+Thumbs.db
+desktop.ini
+
+# ── Environment / Secrets ──
+.env
+.env.*
+!.env.example
+*.key
+*.pem
+bot_credentials.json
+api_key
+
+# ── Claude / AI agent artifacts ──
+.claude/
+.skills/
+.agents/
+
+# ── Browser automation artifacts (page captures, recovery codes, console logs) ──
+.playwright-mcp/
+playwright-report/
+test-results/
+
+# ── Internal knowledge base (Agent WAL, conversations, deployment details) ──
+memall知识库/
+
+# ── Bot scripts (internal, hardcoded IDs) ──
+bot_service.cjs
+debug_token.cjs
+get_token.*
+ws_*
+
+# ── MCP server config (contains API keys) ──
+mcp_servers.json
+
+# ── Desktop tools (internal use only) ──
+desktop/capture_tray.py
+
+# ── Internal ops scripts ──
+scripts/
+start_headroom.bat
+start_mcp_http.ps1
+start_mcp_http.vbs
+sync_to_vault.*
+install.bat
+install.sh
+run_bot.bat
+
+# ── Node.js ──
+node_modules/
+freellmapi/node_modules/
+
+# ── Binary/vendored deps (keep in repo but flag as generated) ──
+sql.js-*.tgz
+
+# ── Database ──
+*.db
+*.db-*
+*.db-shm
+*.db-wal
+*.sqlite
+chroma/
+data/
+chroma_data/
+
+# ── Logs ──
+*.log
+logs/
+bot_runtime.log
+
+# ── Backup / Export ──
+*.bak
+*.bak-*
+exports/
+reports/
+*.csv
+memall-frontend-snapshot.md
+memories-page.md
+
+# ── Build artifacts (frontend bundles live in desktop/v30/) ──
+frontend/dist/
+desktop/v30/
+
+# ── Marketing / sales materials (not source code) ──
+marketing/
+
+# ── Internal / third-party subprojects (not part of MemALL open-source) ──
+freellmapi/
+paperclip/
+.workbuddy/
+
+# ── Test archive (internal debug scripts) ──
+tests/archive/
+_*.py
+_*.txt
+
+# ── Vector model artifacts (regeneratable) ──
+.vector_model/
+
+# ── Internal design docs / plans ──
+*.plan.md
+l5_completion_plan.md
+前端设计文档_v2.md
+
+# ── Marketing media files (large, not code) ──
+marketing/*.mp4
+
+# ── Internal reports / docs ──
+architecture_*.md
+code_review_suggestions.md
+project_recommendations.md
+network_architecture_plan.md
+BOM_list.md
+self_reflection_*.md
+memory_design_reference.md
+recall_result.txt
+tmp_check_debt.py
+
+# ── Board / inventory artifacts ──
+board.html
+board_err.txt
+memall-feature-inventory/
+
+# ── Generated / internal reports (*.docx) ──
+*.docx
+
+# ── Unrelated files accidentally tracked ──
+ODF_jump_table.md
+
+# ── Bridge message inboxes (internal Feishu messages) ──
+src/memall/bridge/inboxes/
+
+# ── Output / test artifacts ──
+output/
+test_frontend.html
+test_mcp_*.jsonl

memall_os-0.1.2/CHANGELOG.md

@@ -0,0 +1,162 @@
+# Changelog
+
+## [v0.1.2] - 2026-06-22
+
+### Added
+
+- **E2E Test Suite**: 25-test end-to-end test covering capture, retrieve, timeline, connect, traverse, session lifecycle, smart store, vector search, DB ops, identity, persona, onboarding, pipeline, index rebuild, dedup, and error handling — all calling `handle_call` directly (no HTTP server) with retry-on-BUSY pattern. (`tests/test_e2e.py`, `tests/test_helpers.py`)
+- **Memory Health System**: New `memall.core.health` module with `collect()` for actionable memory diagnostics. Integrated into `memall doctor --deep` for deep health checks and `session_start` as `[HEALTH]` section. Reports graph coverage, reflection rate, isolated memories, stale discussions, pipeline freshness, and DB size with issue/recommendation hints. (`core/health.py`, `cli/commands/management_commands.py`, `pipeline/session.py`)
+- **Export/Import/Sync System**: JSONL export format with content_hash dedup, `--since` time filter, `memall import <file>` for JSON/JSONL import, and `memall sync --from <file>` for incremental sync with state tracking in `~/.memall/sync_state.json`. (`cli/export.py`, `cli/main.py`, `cli/commands/management_commands.py`)
+
+### Fixed
+
+- **Category Taxonomy Normalization**: Eliminated all 122 composite categories and consolidated 100+ labels → 25 clean categories. Fixed root cause in `integrate.py` (L10 merge no longer concatenates categories with `、`; picks majority category instead). Applied DB cleanup via migration script to standardize synonyms (`bugfix→fix`, `business_idea→business`, `discussion_response→discussion`, `daily_summary→report`, etc.). (`pipeline/integrate.py`)
+- **ops.py SyntaxError**: Moved `import logging` to module level to fix `expected 'except' or 'finally' block` crash introduced in earlier commit. (`pipeline/ops.py`)
+
+### Changed
+
+- **Lazy Auto-Init**: `get_conn()` and `ConnectionPool._new_conn()` now call `init_db()` on their first invocation, so no explicit `memall init` is required for new users or agents that clone the repo. (`core/db.py`)
+
+### Fixed
+
+- **Gateway import global content_hash dedup**: Dedup check was scoped by `agent_name`, but the `UNIQUE` constraint is global — switched to a global lookup. (`gateway.py`)
+- **Connection Pool Write Lock**: `pool_conn()` returned connections with uncommitted implicit write transactions, causing "database is locked" on reused connections. Added `conn.commit()` in pool_conn context manager's finally block. (`core/db.py`)
+- **vec0 Dimension Mismatch**: `build_index()` passed raw k-dim SVD vectors (k ≪ 256 for small datasets) to vec0 expecting 256-dim vectors. Added padding to `EMBED_DIM=256` before `tobytes()`. (`graph/embeddings.py`)
+- **Pipeline Hook TypeError**: `_hook_pipeline_stop` assumed all step results were `int`, but `classify_step()` returns `dict`. Added `_count()` helper to extract integer from dict. (`mcp/hooks_builtin.py`)
+- **OpsInput None Defaults**: Pydantic model had `Optional[int] = None` which `model_dump()` preserved as `None`, causing `TypeError` in dedup operator. Changed to explicit `Field(...)` defaults. (`mcp/models.py`)
+- **`_auto_embed` Missing Table**: Called `SELECT` on `memory_embeddings` before table existed on fresh DB. Added `_ensure_embeddings_table()` guard. (`graph/embeddings.py`)
+- **`_load_embeddings_matrix` Missing Table**: Queried `memory_embeddings` without creating it first. Added `_ensure_embeddings_table()` call. (`graph/embeddings.py`)
+- **`_query_embed` Dimension Mismatch**: SVD produced k-dim query vectors (k < `EMBED_DIM`) causing matmul shape error. Added padding to `EMBED_DIM=256`. (`graph/retrieve.py`)
+- **Migration 015/017/018 Silent Errors**: `logger = logging.getLogger(__name__)` placed inside docstrings, never executed — migrations silently caught all exceptions. Extracted logger assignment above docstring. (`migrations/015_*.py`, `migrations/017_*.py`, `migrations/018_*.py`)
+- **Missing `identity_profile` Column**: Column referenced in code but missing from base schema DDL. Added to `CREATE TABLE identities`. (`core/db.py`)
+- **Thread-Safe Connection Close**: `ConnectionPool.get()` tried to close connections owned by another thread, causing `ProgrammingError`. Added specific catch for `sqlite3.ProgrammingError`. (`core/db.py`)
+- **SyntaxWarning `\\w`**: Invalid escape sequence `\w` in docstring triggered Python 3.12 warning. Escaped backslash. (`graph/embeddings.py`)
+- **`doctor --deep` UnboundLocalError**: Redundant `import json` inside `cmd_doctor()` shadowed the module-level import, causing `UnboundLocalError` on all non-`--fix` runs. Removed the local import. (`cli/commands/management_commands.py`)
+- **MCP stdout GBK crash**: `_respond()` wrote JSON with `ensure_ascii=False` to `sys.stdout`, which crashes on Windows GBK consoles when Unicode chars (✅) appear. Added `sys.stdout.reconfigure(encoding='utf-8')` at `serve()` entry + `PYTHONIOENCODING=utf-8` env var in MCP config. (`mcp/server.py`, `.claude/settings.json`)
+- **DB default on C: drive**: `_resolve_db_path()` now prefers first available non-system drive (D:, E:, …) on Windows instead of always dropping in `C:\Users\...\.memall`. Backups and `memall doctor` path checks follow the same logic. (`core/db.py`, `cli/backup_restore.py`, `cli/commands/management_commands.py`)
+
+## [v0.1.1] - 2026-06-21
+
+### Fixed
+
+- **HTTP Transport Crash**: Root cause fixed — sync `handle_call()` blocked aiohttp event loop. Offloaded to `ThreadPoolExecutor` (12 fast + 2 heavy workers) with `asyncio.wait_for()`. Auto-restart on crash/port conflict. (`http_transport.py`, `shared.py`)
+- **DB Connection Deadlock**: `ConnectionPool.get()` had no timeout on `Queue.get()` — added 30s barrier. 21 raw `sqlite3.connect()` calls missing `timeout=10` — all backfilled across federation, lark, cli, pipeline modules. (`core/db.py`, 8 federation/cli/api files)
+
+### Added
+
+- **L7 Lifecycle Closure**: `auto_inject` defaults to True across all entry points (5 files) — new sessions automatically inject `[L7约束]` behavioral rules. L6→L7 auto-distillation via `distill_l7.py` regex-based lesson extraction, registered in pipeline after `reflect_step()`. (`pipeline/distill_l7.py`, `mcp/models.py`, `mcp/tools/__init__.py`, `mcp/tools/session.py`, `pipeline/session.py`, `api/server.py`, `pipeline/pipeline.py`)
+
+### Changed
+
+- **CLAUDE.md**: Added "自动提交" rule — each independent change auto-updates ALL relevant .md (not just CHANGELOG) + commit + push + notify user.
+
+### Changed
+
+- **Lazy Auto-Init**: `get_conn()` and `ConnectionPool._new_conn()` now call `init_db()` on their first invocation, so no explicit `memall init` is required for new users or agents that clone the repo. (`core/db.py`)
+
+## [v0.1.0] - 2026-06-19
+
+### Added
+
+- **Memory Lifecycle**: 10-layer memory architecture (P0/L1-L10) with automatic pipeline
+- **Decision Arcs**: Full L4→L5→L6 lifecycle with convergence engine for multi-agent discussions
+- **Timeline System**: Pre-aggregated time_slices (day/week/month) + epoch detection (gaps, topic drift, reflection inflection points)
+- **Self-Reflection (L6)**: Automatic quality review, pattern recognition, error correction
+- **Knowledge Distillation (L9)**: Compress raw memories into structured knowledge graph
+- **Multi-Agent Federation**: Cross-agent memory publish/query/conflict resolution with trust hierarchy
+- **LAN Discovery**: Auto-detect nearby peers via mDNS, bidirectional sync
+- **Hybrid Search**: FTS5 exact match + sqlite-vec (256-dim) semantic similarity
+- **Session Management**: session_start with auto-inject, session_end with summary, session_summary
+- **Agent Identity**: L1 identity traits + L7 preferences profiling
+- **Onboarding System**: 5-step guided setup for new users
+- **OODA Self-Improvement**: Observe-Orient-Decide-Act loop without human intervention
+- **Quality Gates**: 8-dimension scoring in pipeline (relevance, coherence, novelty, actionability, etc.)
+- **Auto-Forget**: TTL expiration + low-value decay with review mechanism
+- **Memory Ops**: merge, split, tag, archive, restore, dedup tools
+- **Security Governance**: audit, permit, check, score subsystem
+- **Gateway Server**: HTTP export/import, LAN discovery, federated queries
+
+### Changed
+
+- **Architecture Redesign**: From legacy 62-action surface to Thin Waist 5-method (capture/recall/connect/traverse/timeline)
+- **MCP Tool Consolidation**: 19 independent tools → 4 tool sets (core, AI, graph, system) → 37 unified MCP tools
+- **Pipeline v3**: 21-step automatic pipeline (enrich → classify → time_slice → arc_status → echo → epoch → reflect → distill → integrate → ...)
+- **Configuration**: All config stored in SQLite `config` table, env overridable
+- **Identity evolved**: Agent identities table with L1/L7 portrait generation
+- **Discipline Migration**: Legacy daemon → Windows Scheduled Tasks (04:00 pipeline, 03:00 forget)
+- **MCP Server**: Unified STDIO + HTTP transport via config-based routing
+- **Pricing positioning**: Freemium model (Free: 5k memory limit, Pro: $9.99/mo) defined
+
+### Fixed
+
+- **DB Path Resolution**: Config-based path respecting overrides (#7905, #8144)
+- **OpenBLAS OOM**: Pipeline crash on 2000+ memories (#8133)
+- **Scheduler Restored**: After 12-day downtime, migrated to Windows Tasks (#7895)
+- **Discussion Dual-Path**: _meta/value duplicate entries (#8177)
+- **Discussion Metadata Migration**: Legacy table drop without re-wrap cycle (#7958, #7963)
+- **Silent Errors**: 79 blocks across 33 files migrated from bare pass to logger.warning (#7965)
+- **Database Copy Bug**: Fixed concurrent write corruption (#5558)
+- **Classify Level Loss**: layer field not persisted in classify step (#4894)
+- **Bridge N+1**: Per-edge queries converted to batch IN (#5306)
+- **Migration Cleanup**: Double migration system removed (#5305)
+- **Test Isolation**: conftest.py + production DB protection (#6292)
+- **10-Layer Health Skew**: Resolved architecture imbalance (#4975, #4977)
+- **Consumer Recovery**: Message consumption restored after refactor (#4979-#4981)
+- **FTS5 Repair**: memory-doctor.py for database integrity checks
+- **Discussion Status**: Removed bare status after cleanup (#7963)
+- **Backup Restoration**: Added memall backup/restore/check commands
+
+### Removed
+
+- Legacy daemon process (replaced by Windows Scheduled Tasks)
+- _run_migrations + 7 migration files (dual migration system)
+- Legacy SmartMemoryInjector (integrated into pipeline)
+- Kronvex from comparison table (blocking marketplace listing)
+- FTS5 as standalone MCP tool (SQLite built-in, not a tool)
+
+### Security
+
+- **3-Layer Safety Net**: Permission + circuit-breaker + recovery
+- **QR Pairing**: LAN device authentication without network exposure
+- **PII Redaction**: Optional content sanitization in scrape/parse pipelines
+- **API Key Auth**: MCP Server authentication module
+
+## [v0.0.2] - 2026-06-06
+
+### Added
+
+- Phase 2 compression and decay mechanisms (DreamGenerator, MemoryLifecycle)
+- Timeline dimension: time_slices, epochs, session summary injection
+- Decision Arc: full L4→L5→L6 lifecycle
+- Discussion convergence engine
+- MCP Marketplace listing draft
+- LAN discovery and federation prototype
+
+### Changed
+
+- Legacy -> Thin Waist architecture migration completed
+- 19 MCP tools -> 4 tool sets
+- Tag normalization: 352 unique tags → 33 (91% reduction), 5-dimension standard set
+- Database path: sandbox (~/.MemALL) → workspace
+
+### Fixed
+
+- Dead code and script cleanup
+- L9 decay pipeline timeout
+- FTS5 + vector search hybrid
+- BOM illegal characters (8 files)
+- Indentation/syntax errors (5 files)
+
+## [v0.0.1] - 2026-05-25
+
+### Added
+
+- Initial MemALL prototype with SQLite-backed memory storage
+- MCP server with HTTP + STDIO transport
+- CLI with 40+ subcommands
+- Agent SDK Python client
+- Basic capture/retrieve/timeline/search operations
+- 10-layer architecture: P0/L1-L10
+- FTS5 full-text search
+- Identity and Agent Registry
+- Self-improvement framework (HOT memory injection)

memall_os-0.1.2/CLAUDE.md

@@ -0,0 +1,28 @@
+# MemALL 开发约束
+
+## 自动存入记忆
+每次完成一个独立改动（修复、重构、新增功能）后，**自主**调用 `capture()` 存入 summary 记忆，不需要等待用户提醒。包括：
+- 代码改动完成后立即存
+- 重要分析/决策结论存
+- session 反思自主存（L6）
+
+## 修复流程
+1. 先分析根因，再考虑 workaround
+2. 异常数据分布先查数据流路径的 WHERE 条件
+3. 改完后做完整 import/parse + 干运行验证
+
+## 自我反思
+1. 每个 session 结束时，自主做一次 L6 反思：做对了什么、做错了什么、改进点
+2. 反思存入数据库（level=L6, category=reflection），不等用户要求
+
+## 自我改进
+1. L1/L7 教训：遇到数据分布异常先查 WHERE 条件链，不凭直觉下结论
+2. 修复节奏：改完做完整验证（import + parse + 干运行），不分步 debug
+3. 根因优先：改一行 SQL → 改配置 → 加新模块，顺序不可倒置
+
+## 自动提交
+每次完成一个独立改动后（修复/重构/新增功能 + 验证通过）：
+1. **更新文档**：在 `CHANGELOG.md` 末尾追加条目（日期 + 摘要 + 涉及文件），同时排查所有相关 `.md`（`README*.md`、`QUICKSTART.md`、`architecture_*.md`、`COMPARISON.md` 等），按实际情况更新内容——不局限于追加，过期内容要改、废弃内容要删
+2. **commit + push**：`git add -A && git commit -m "type: summary..." && git push`
+3. commit message 用英文，格式：`type: description`（type: fix/feat/refactor/docs/chore）
+4. 推送完后主动告知用户（一句短消息，例如 "已推送 (hash)"）

memall_os-0.1.2/COMPARISON.md

@@ -0,0 +1,71 @@
+# 竞品对比：MemALL vs Mem0 / Letta / Zep
+
+## 核心定位差异
+
+| 维度 | MemALL | Mem0 | Letta | Zep |
+|------|--------|------|-------|-----|
+| **定位** | Multi-agent Memory OS / Loop Engineering 记忆基础设施 | 个人 AI 记忆层 | Agent 服务端框架 | 对话记忆中间件 |
+| **记忆模型** | 10 层生命周期（P0-L10） | 用户/会话 两层 | 智能体/记忆块 | 会话/摘要 两层 |
+| **持久化** | SQLite（本地优先） | 云 API | PostgreSQL | 云 API |
+| **协议** | MCP Server | REST API | REST + gRPC | REST API |
+| **开源** | ✅ 全开源 | ⚠️ 部分 | ✅ | ⚠️ 部分 |
+
+## 功能差异
+
+### 记忆生命周期
+
+MemALL 独有：P0（紧急）→ P1/P2（规划）→ L1-L10（认知层级），每条记忆有明确的层级语义和转换规则。竞品只有"用户记忆"和"会话记忆"两类。
+
+### 时间线
+
+| 能力 | MemALL | Mem0 | Letta | Zep |
+|------|--------|------|-------|-----|
+| 预聚合时间片 | time_slices（日/周/月） | ❌ | ❌ | ❌ |
+| 时期分段 | epochs（自动检测 gap/主题漂移/反思拐点） | ❌ | ❌ | ❌ |
+| 时间衰减 | temporal_weight（指数+epoch_boost） | ❌ | ❌ | 基础衰减 |
+| 感知时长 | ❌（Phase 3） | ❌ | ❌ | ❌ |
+
+### 决策追踪
+
+| 能力 | MemALL | Mem0 | Letta | Zep |
+|------|--------|------|-------|-----|
+| 决策生命周期 | 决策弧（open→in_progress→closed） | ❌ | ❌ | ❌ |
+| 任务关联 | L5 任务自动关联决策 | ❌ | ❌ | ❌ |
+| 反思闭环 | L6 反思自动闭合决策弧 | ❌ | ❌ | ❌ |
+| 未闭合决策注入 | session_start 自动提示 | ❌ | ❌ | ❌ |
+
+### 记忆管线
+
+MemALL 有 21 步自动管线（enrich → classify → time_slice → arc_status → echo → epoch → reflect → distill → integrate → ...）。竞品依赖用户手动触发或简单 CRUD。
+
+### 讨论收敛
+
+MemALL Phase 1.5 支持多 Agent 讨论自动推进到共识并 capture 为可追溯的决策记忆。竞品无此能力。
+
+### 多 Agent 协作
+
+| 能力 | MemALL | Mem0 | Letta | Zep |
+|------|--------|------|-------|-----|
+| 跨 Agent 共享 | 共享记忆 + 权限控制 | ❌ | 同 Agent 内 | ❌ |
+| Agent 身份 | identities 表 + L1/L7 画像 | ❌ | 智能体定义 | ❌ |
+| 联邦记忆 | 跨设备同步 + LAN 发现 | ❌ | ❌ | ❌ |
+
+## 适用场景
+
+| 场景 | 推荐 |
+|------|------|
+| 个人 AI 助手记忆 | Mem0 / MemALL |
+| 单 Agent 长期对话 | Zep |
+| 自主 Agent 服务端 | Letta |
+| **多 Agent 协作系统** | **MemALL** |
+| **Loop Engineering 基础设施** | **MemALL** |
+| **本地优先 / 离线场景** | **MemALL** |
+| 团队 / 企业级 | Letta / Zep Cloud |
+
+## MemALL 的核心差异总结
+
+1. **生命周期深度**：10 层不是噱头，每条记忆知道自己的"认知权重"
+2. **时间线维度**：time_slices + epochs = Agent 知道自己"从哪来、在哪个阶段"
+3. **决策弧**：让 Agent 知道"哪些决策没有收尾"，这是 Loop Engineering 中 Sub-agents 协作的前提
+4. **离线优先**：SQLite 本地运行，不依赖云 API，适合需要数据主权的场景
+5. **管线自动化**：21 步管线自动处理，无需手动维护

memall_os-0.1.2/CONTRIBUTING.md

@@ -0,0 +1,42 @@
+# Contributing to MemALL
+
+Thanks for your interest in contributing! We're an open-source, community-driven project and welcome contributions of all kinds.
+
+## 🐛 Bug Reports
+
+Open a [GitHub Issue](https://github.com/j19800/MemALL/issues/new) with:
+- A clear title and description
+- Steps to reproduce
+- Expected vs actual behavior
+- Your environment (OS, Python version, MemALL version)
+
+## 💡 Feature Requests
+
+Open a [Discussion](https://github.com/j19800/MemALL/discussions) first. We use L4 decisions internally to track feature planning — your proposal will be treated the same way.
+
+## 🔧 Pull Requests
+
+1. Fork the repo
+2. Create a feature branch (`git checkout -b feat/your-feature`)
+3. Make your changes
+4. Run tests: `pytest`
+5. Open a PR against `main`
+
+### PR Checklist
+- [ ] Tests pass (`pytest`)
+- [ ] New tests added if applicable
+- [ ] Code follows existing style (we use `black` + `ruff`)
+- [ ] Commit messages are clear
+
+## 🧪 Development Setup
+
+```bash
+git clone https://github.com/j19800/MemALL
+cd memall
+pip install -e .[dev]
+pytest
+```
+
+## 📋 Code of Conduct
+
+Be respectful. We're building something cool — let's keep it fun.

memall_os-0.1.2/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024-2025 MemALL Contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.