Memory Keeper

Copy and snapshot all important agent context (MEMORY.md, memory/*.md, AGENTS.md, SOUL.md, etc.) into a dedicated archive directory or repo. Use this skill when you want to back up your memories, context, or configuration files in preparation for maintenance, corruption recovery, or transferring to another host.

This skill appears to do exactly what it claims: back up your agent memory files and optionally commit/push them to a git repo. Before installing or running it: (1) ensure you have git installed if you plan to use --commit/--push; (2) set the --target to a local path you control and review the archive after a dry run to confirm only intended files were copied; (3) be careful with --allow-extra — it can include any file/glob from your workspace (avoid adding secrets, keys, or tokens); (4) when using --remote/--push, only push to repositories you control — embedding credentials in remote URLs or pushing to unknown remotes can exfiltrate data; (5) note the tool appends a timestamped entry to memory/YYYY-MM-DD.md (it modifies your workspace logs) — if you want a read-only snapshot, inspect the code or run it against a copied workspace first. If you want higher assurance, review the bundled scripts (memory_sync.py) and run the unit tests locally.

Type: OpenClaw Skill Name: Developer: Version: Description: OpenClaw Agent Skill Suspicious High-Entropy/Eval files: 2 The `scripts/memory_sync.py` skill is classified as suspicious due to its broad file access and network capabilities. Specifically, the `--allow-extra` argument allows copying arbitrary files or directories from the agent's workspace (or paths accessible via globs from the workspace) to a target archive. This, combined with the ability to configure an arbitrary git remote URL and push all committed changes to it, creates a significant risk of data exfiltration if the skill is invoked with malicious parameters. These capabilities are explicitly documented in `SKILL.md`, `README.md`, and `references/usage.md` as features, but they lack sufficient safeguards to prevent misuse for exfiltrating sensitive data beyond the stated purpose of backing up agent context.