Simplemem

Efficient Lifelong Memory for LLM Agents - semantic compression, cross-session memory, and intent-aware retrieval

This skill appears to do what it claims, but consider the following before installing: 1) The wrapper will use your OPENAI_API_KEY to call external APIs (billing and data-sharing implications) — consider using a dedicated key or billing alerts. 2) SKILL.md recommends 'pip install simplemem' or cloning GitHub; review the upstream PyPI package and/or repository code before installing, because installing a pip package runs code from that package. 3) The manifest mentions a PowerShell CLI (simplemem.ps1) in docs but that file is not included — expect the included Python wrapper to be the provided functionality. 4) Memories are persisted locally under the skill's data directory (data/memories.json and possibly a lancedb); if this contains sensitive data, store it securely or clear it as needed. 5) If you need higher assurance, audit the upstream project (https://github.com/aiming-lab/SimpleMem) or run installations in an isolated environment (container/VM) first.

Type: OpenClaw Skill Name: simplemem Version: 1.0.1 The skill bundle provides a wrapper for a semantic memory system, SimpleMem. The `SKILL.md` file contains standard installation and usage instructions, with no evidence of prompt injection attempts against the agent. The `simplemem.py` script handles local data storage within its own `data` directory and uses the `OPENAI_API_KEY` for its stated purpose of interacting with the OpenAI API for semantic features. There is no indication of malicious activities such as data exfiltration, unauthorized command execution, persistence mechanisms, or obfuscation. All observed behaviors are aligned with the skill's described functionality.