功能描述

Curate Claude Code's auto-memory into durable project knowledge. Analyze MEMORY.md for patterns, promote proven learnings to CLAUDE.md and .claude/rules/, ex...

使用说明 (SKILL.md)

Self-Improving Agent

Name: Auto Memory Pro
Author: alirezarezvani

Auto-memory captures. This plugin curates.

Claude Code's auto-memory (v2.1.32+) automatically records project patterns, debugging insights, and your preferences in MEMORY.md. This plugin adds the intelligence layer: it analyzes what Claude has learned, promotes proven patterns into project rules, and extracts recurring solutions into reusable skills.

Quick Reference

Command	What it does
`/si:review`	Analyze MEMORY.md — find promotion candidates, stale entries, consolidation opportunities
`/si:promote`	Graduate a pattern from MEMORY.md → CLAUDE.md or `.claude/rules/`
`/si:extract`	Turn a proven pattern into a standalone skill
`/si:status`	Memory health dashboard — line counts, topic files, recommendations
`/si:remember`	Explicitly save important knowledge to auto-memory

How It Fits Together

┌─────────────────────────────────────────────────────────┐
│                  Claude Code Memory Stack                │
├─────────────┬──────────────────┬────────────────────────┤
│  CLAUDE.md  │   Auto Memory    │   Session Memory       │
│  (you write)│   (Claude writes)│   (Claude writes)      │
│  Rules &    │   MEMORY.md      │   Conversation logs    │
│  standards  │   + topic files  │   + continuity         │
│  Full load  │   First 200 lines│   Contextual load      │
├─────────────┴──────────────────┴────────────────────────┤
│              ↑ /si:promote        ↑ /si:review          │
│         Self-Improving Agent (this plugin)               │
│              ↓ /si:extract    ↓ /si:remember            │
├─────────────────────────────────────────────────────────┤
│  .claude/rules/    │    New Skills    │   Error Logs     │
│  (scoped rules)    │    (extracted)   │   (auto-captured)│
└─────────────────────────────────────────────────────────┘

Installation

Claude Code (Plugin)

/plugin marketplace add alirezarezvani/claude-skills
/plugin install self-improving-agent@claude-code-skills

OpenClaw

clawhub install self-improving-agent

Codex CLI

./scripts/codex-install.sh --skill self-improving-agent

Memory Architecture

Where things live

File	Who writes	Scope	Loaded
`./CLAUDE.md`	You (+ `/si:promote`)	Project rules	Full file, every session
`~/.claude/CLAUDE.md`	You	Global preferences	Full file, every session
`~/.claude/projects/\x3Cpath>/memory/MEMORY.md`	Claude (auto)	Project learnings	First 200 lines
`~/.claude/projects/\x3Cpath>/memory/*.md`	Claude (overflow)	Topic-specific notes	On demand
`.claude/rules/*.md`	You (+ `/si:promote`)	Scoped rules	When matching files open

The promotion lifecycle

1. Claude discovers pattern → auto-memory (MEMORY.md)
2. Pattern recurs 2-3x → /si:review flags it as promotion candidate
3. You approve → /si:promote graduates it to CLAUDE.md or rules/
4. Pattern becomes an enforced rule, not just a note
5. MEMORY.md entry removed → frees space for new learnings

Core Concepts

Auto-memory is capture, not curation

Auto-memory is excellent at recording what Claude learns. But it has no judgment about:

Which learnings are temporary vs. permanent
Which patterns should become enforced rules
When the 200-line limit is wasting space on stale entries
Which solutions are good enough to become reusable skills

That's what this plugin does.

Promotion = graduation

When you promote a learning, it moves from Claude's scratchpad (MEMORY.md) to your project's rule system (CLAUDE.md or .claude/rules/). The difference matters:

MEMORY.md: "I noticed this project uses pnpm" (background context)
CLAUDE.md: "Use pnpm, not npm" (enforced instruction)

Promoted rules have higher priority and load in full (not truncated at 200 lines).

Rules directory for scoped knowledge

Not everything belongs in CLAUDE.md. Use .claude/rules/ for patterns that only apply to specific file types:

# .claude/rules/api-testing.md
---
paths:
  - "src/api/**/*.test.ts"
  - "tests/api/**/*"
---
- Use supertest for API endpoint testing
- Mock external services with msw
- Always test error responses, not just happy paths

This loads only when Claude works with API test files — zero overhead otherwise.

Agents

memory-analyst

Analyzes MEMORY.md and topic files to identify:

Entries that recur across sessions (promotion candidates)
Stale entries referencing deleted files or old patterns
Related entries that should be consolidated
Gaps between what MEMORY.md knows and what CLAUDE.md enforces

skill-extractor

Takes a proven pattern and generates a complete skill:

SKILL.md with proper frontmatter
Reference documentation
Examples and edge cases
Ready for /plugin install or clawhub publish

Hooks

error-capture (PostToolUse → Bash)

Monitors command output for errors. When detected, appends a structured entry to auto-memory with:

The command that failed
Error output (truncated)
Timestamp and context
Suggested category

Token overhead: Zero on success. ~30 tokens only when an error is detected.

Platform Support

Platform	Memory System	Plugin Works?
Claude Code	Auto-memory (MEMORY.md)	✅ Full support
OpenClaw	workspace/MEMORY.md	✅ Adapted (reads workspace memory)
Codex CLI	AGENTS.md	✅ Adapted (reads AGENTS.md patterns)
GitHub Copilot	`.github/copilot-instructions.md`	⚠️ Manual promotion only

This package appears to do what it says: it reads your auto-memory, identifies patterns, and can promote or extract them into rules or skills. Key things to check before enabling: - Hooks behavior: README and other docs say the error-capture hook "appends a structured entry to auto-memory," but the shipped hook script only prints a <error-detected> block to stdout (it does not append to MEMORY.md). Confirm whether your platform captures that output into memory automatically or whether the hook only notifies you and expects a manual /si:remember. Don't assume it writes to files. - File writes on promote/extract: /si:promote and /si:extract can create or modify local files (CLAUDE.md, .claude/rules/, skills/*) and remove entries from MEMORY.md. Make a backup of your memory/rules files or review changes interactively before allowing automatic writes. - Local file access is broad but relevant: the skill reads ~/.claude/projects/... and project files. That's necessary for its function, but be comfortable with it accessing those paths. - No network downloads or secret requests were found. Still, test in a safe environment (or with a dry-run option) to confirm the actual side effects match the documented behavior. If you want to proceed: run /si:review or /si:status in a dry-run mode first, inspect the proposed edits, and enable the hook only after confirming how your platform handles hook output.

功能分析

Type: OpenClaw Skill Name: auto-memory-pro Version: 1.0.0 The bundle implements a high-risk PostToolUse hook (hooks/error-capture.sh) that monitors all bash command output (CLAUDE_TOOL_OUTPUT) to capture errors into the agent's memory, which could inadvertently expose sensitive data or secrets printed to the console. Additionally, the skill's core functionality involves modifying the agent's primary instruction files (CLAUDE.md and .claude/rules/), providing a mechanism for persistent behavior modification. While these capabilities are aligned with the stated purpose of curating 'auto-memory,' the combination of broad file system access to ~/.claude/projects/ and the monitoring of tool outputs represents a significant risk profile.

能力评估

ℹ Purpose & Capability

Name/description match what the files do: analyze MEMORY.md, propose/promote rules, and extract portable skills. The included agents (memory-analyst, skill-extractor) and promote/remember/extract/review/status commands are coherent with that purpose. Minor note: some documentation claims the hook 'appends a structured entry to auto-memory' automatically, but the provided hook script only prints a structured message (does not append).

⚠ Instruction Scope

Runtime instructions explicitly tell the agent to read local memory directories (e.g. ~/.claude/projects/.../memory/, ./CLAUDE.md, .claude/rules/) and to run shell commands like grep, ls, find. That is appropriate for a local memory-curation tool, but it does mean the skill will read arbitrary files under the declared memory paths and the repository. The discrepancy between claimed automatic write-to-memory behavior and the hook script's actual behavior (prints guidance instead of writing) is an incoherence you should be aware of.

✓ Install Mechanism

No install spec and no remote downloads: instruction-only with a small local hook script. This is low-risk from an installation standpoint because nothing is fetched from external URLs.

✓ Credentials

The skill requires no credentials or external environment variables. The hook script expects CLAUDE_TOOL_OUTPUT in its runtime environment (provided by the host during PostToolUse), which is reasonable for a hook but is not declared as a required env var in metadata — a minor documentation gap rather than a secret-exfiltration signal.

ℹ Persistence & Privilege

always:false (normal). The skill includes actions that can write to project files (promote writes to CLAUDE.md or .claude/rules/ and can remove entries from MEMORY.md) — this is expected for the feature but is a write-level privilege to your repository and local Claude config, so you should explicitly confirm file writes before use.

版本历史

v1.0.0

auto-memory-pro v1.0.0 – Initial Release - Introduces a plugin that curates Claude Code's auto-memory (MEMORY.md), promoting key patterns to enforceable project rules and extracting recurring solutions as reusable skills. - New commands: review memory for promo candidates, promote patterns, extract skills, check memory health, explicitly save knowledge. - Provides agents for memory analysis and skill extraction. - Adds error-capture hook to auto-record CLI issues into project memory. - Documentation included for workflows, memory architecture, platform support, and use cases.

元数据

Slug auto-memory-pro

版本 1.0.0

许可证 —

累计安装 1

当前安装数 1

历史版本数 1

常见问题

Auto Memory Pro 是什么？

Curate Claude Code's auto-memory into durable project knowledge. Analyze MEMORY.md for patterns, promote proven learnings to CLAUDE.md and .claude/rules/, ex... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 365 次。

如何安装 Auto Memory Pro？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install auto-memory-pro」即可一键安装，无需额外配置。

Auto Memory Pro 是免费的吗？

是的，Auto Memory Pro 完全免费（开源免费），可自由下载、安装和使用。

Auto Memory Pro 支持哪些平台？

Auto Memory Pro 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Auto Memory Pro？

由 Alireza Rezvani（@alirezarezvani）开发并维护，当前版本 v1.0.0。

Auto Memory Pro