Memory Hamster

Agent 记忆进化系统 - 温度模型 + 自动归档 + 学习记录 + 技能提炼。让 AI 每天变得更聪明。

This skill appears to implement exactly what it says: local memory management, archiving, nightly reflections, a local semantic search, and a helper to scaffold skills. Before installing or enabling cron jobs: - Review the scripts (memory-gc.sh, nightly-reflection.sh, extract-skill.sh, search-memory.cjs) to confirm the target WORKSPACE path is correct so they don't move files you care about. - Run scripts in dry-run modes where provided (memory-gc.sh and extract-skill.sh support dry-run) or run them manually first to observe behavior. - Back up your existing workspace/memory/ files if you have important notes; GC will move >30-day .md files into .archive (not deleted but moved). - search-memory.cjs requires Node (package.json requests node >=16); ensure runtime compatibility. - The package metadata references external hosts but the skill's published source/homepage are unclear—if provenance matters to you, obtain the upstream repository or author verification before enabling automated runs. Overall this is internally consistent and does not request secrets or network access, but it performs filesystem changes so review and test locally before making it run automatically.

Type: OpenClaw Skill Name: memory-hamster Version: 1.0.1 The memory-hamster bundle is a memory management and self-evolution system for AI agents. It provides scripts (memory-gc.sh, nightly-reflection.sh, search-memory.cjs) to archive old logs, generate daily reflections, and perform keyword-based searches across the agent's workspace. The scripts use standard system utilities and include basic security checks, such as path traversal validation in extract-skill.sh, and there is no evidence of data exfiltration, malicious execution, or harmful prompt injection.