Monica Memory Manager
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: monica-memory-manager Version: 1.0.0 The OpenClaw AgentSkills bundle 'monica-memory-manager' is classified as benign. The skill's purpose is to manage a three-tier memory system (short, medium, long term) for an AI agent, which is implemented using local file operations and a local ChromaDB vector store within the designated OpenClaw workspace. The `scripts/memory_manager.py` script uses `argparse` for safe command-line argument handling and `pathlib` for secure file path manipulation. While the `generate_summary` function contains a `TODO: 接入实际 LLM` comment, its current implementation is a harmless placeholder that does not make external API calls. The documentation (`SKILL.md` and `references/references.md`) provides clear instructions and safe `subprocess.run` examples, with no evidence of prompt injection attempts or other malicious instructions. There are no indicators of data exfiltration, persistence mechanisms, unauthorized network activity, or obfuscation.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Information saved as memory may persist beyond the current conversation and influence later responses.
The skill explicitly creates long-term persistent memory for later retrieval, which is central to its purpose but means personal or sensitive conversation content may be reused in future context.
长期 | `memory/vector-store/` | 语义检索 | 永久记忆，RAG
Only store content you are comfortable keeping as long-term memory, and review the local memory directory periodically.
Memories could be written somewhere unexpected if WORKSPACE_DIR is not set.
The script stores memory under WORKSPACE_DIR but otherwise falls back to a user-specific absolute path, so users should verify where persistent memory files will be written.
WORKSPACE_DIR = Path(os.environ.get('WORKSPACE_DIR', '/Users/scott/.openclaw/workspace'))Set WORKSPACE_DIR explicitly before use, or adjust the script/configuration so memory is stored in the intended project or OpenClaw workspace.
Installing the dependency manually may introduce version drift or package-source risk.
The long-term vector-store feature relies on ChromaDB, but the skill has no install spec or pinned dependency in the provided artifacts.
print("✗ 需要安装 chromadb: pip install chromadb")Install ChromaDB from a trusted package index, consider pinning a known-good version, and review dependency changes before enabling long-term memory.