← Back to Skills Marketplace
136
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install agent-memory-ops
Description
Audit and maintain OpenClaw-style long-term memory. Use for MEMORY.md cleanup, daily-note digestion, duplicate detection, stale-memory review, and promoting...
README (SKILL.md)
Agent Memory Ops
Use this skill when you need to keep an agent's memory layer healthy instead of letting MEMORY.md rot.
What it does
- scans
MEMORY.md+memory/*.md - detects duplicate / near-duplicate bullets
- extracts memory candidates from recent daily notes
- surfaces active follow-ups / TODOs
- filters obvious secrets from suggested memory output
- produces a concise maintenance report you can act on
Good use cases
- "帮我整理 MEMORY.md"
- "检查记忆层有没有重复和过期信息"
- "把最近几天的重要内容沉淀到长期记忆"
- "做一次 agent memory audit"
- "维护长期记忆 / daily memory / notebook memory"
Commands
Run from the workspace root that contains MEMORY.md and memory/.
python3 {baseDir}/scripts/memory_ops.py report --root .
python3 {baseDir}/scripts/memory_ops.py dedupe --root .
python3 {baseDir}/scripts/memory_ops.py digest --root . --days 7
python3 {baseDir}/scripts/memory_ops.py digest --root . --files 5 --format json
Recommended workflow
- Run
reportto see gaps, duplicates, and follow-ups. - Run
digeston the last 5-7 daily notes. - Promote only durable facts into
MEMORY.md. - Keep volatile chatter in daily notes.
- Never copy secrets into curated memory unless the user explicitly asks.
Output policy
- Prefer
--format markdownfor human review. - Prefer
--format jsonwhen another tool or script will consume the result. - The script intentionally redacts / skips likely secrets from digest suggestions.
References
references/playbook.md
Usage Guidance
This skill appears coherent and implements the advertised memory-audit functionality, but take these precautions before running it: 1) Inspect the full scripts/memory_ops.py file yourself (the prompt included a truncated view) to confirm there are no hidden network calls or surprising behavior. 2) Run the script locally in a safe/dev environment first — it reads and prints items from MEMORY.md and memory/*.md, so outputs could expose sensitive content. 3) Don't run it against repos containing live secrets; the secret-detection is heuristic and can miss things. 4) Back up MEMORY.md before applying any automated changes and manually review suggestions before promoting items into curated memory.
Capability Analysis
Type: OpenClaw Skill
Name: agent-memory-ops
Version: 0.1.1
The agent-memory-ops skill bundle is designed to manage and audit local markdown-based memory files. The core logic in scripts/memory_ops.py focuses on deduplication and categorization of text using standard libraries (difflib, re, pathlib) and includes security-conscious features such as regex-based filtering to prevent the accidental promotion of secrets (API keys, tokens) into long-term memory. No network activity, file-writing operations, or malicious prompt-injection attempts were found.
Capability Assessment
Purpose & Capability
Name/description match what is present: SKILL.md and scripts/memory_ops.py implement scanning MEMORY.md and memory/*.md, deduping, digesting, and reporting. No unrelated binaries, credentials, or config paths are requested.
Instruction Scope
Instructions correctly tell the agent to run the included Python script from the workspace root. The script reads MEMORY.md and files under memory/*.md and prints markdown/JSON reports including extracted bullet text. This is within scope, but note that the tool reads and outputs content from local files (which may contain sensitive data). The script contains heuristics to filter likely secrets but uses pattern matching (heuristic), so it may miss some secrets or redact too little — review outputs before promoting content.
Install Mechanism
No install spec is provided (instruction-only skill) and the repository includes a Python script to run. Nothing is downloaded or written to disk by an installer; running the script executes only local code.
Credentials
The skill declares no required environment variables, no primary credential, and no config paths. The script likewise does not read environment variables or external credentials. Requested access is proportional to the stated task (reading local memory files).
Persistence & Privilege
always:false and default autonomous-invocation are set (normal). The skill does not request permanent system-wide privileges or modify other skills' configs per the provided files.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agent-memory-ops - After installation, invoke the skill by name or use
/agent-memory-ops - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.1
Trimmed the skill package, removed redundant docs, regenerated agent metadata, and revalidated the memory audit workflow on a real OpenClaw workspace.
v0.1.0
Initial release: memory audit, dedupe detection, and digest generation for MEMORY.md + daily notes.
Metadata
Frequently Asked Questions
What is Agent Memory Ops?
Audit and maintain OpenClaw-style long-term memory. Use for MEMORY.md cleanup, daily-note digestion, duplicate detection, stale-memory review, and promoting... It is an AI Agent Skill for Claude Code / OpenClaw, with 136 downloads so far.
How do I install Agent Memory Ops?
Run "/install agent-memory-ops" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Memory Ops free?
Yes, Agent Memory Ops is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Agent Memory Ops support?
Agent Memory Ops is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Memory Ops?
It is built and maintained by Orime (@orime); the current version is v0.1.1.
More Skills