Memscape

Persistent memory and collective knowledge for AI agents. Use at session start to load previous context (memscape_resume). Use during work to save decisions,...

This skill appears coherent for a remote memory/knowledge service, but treat it like any external data store: 1) keep MEMSCAPE_API_KEY secret and store it only in the indicated credentials file; 2) avoid saving passwords, private keys, access tokens, or other secrets in 'memories' you promote or contribute; 3) review memscape.org's privacy/retention policy before storing project data (memories may persist and be shared when promoted); 4) when following host-specific setup steps, double-check you are editing the intended config files (back them up first); 5) if you need stricter guarantees, consider restricting the agent from using memscape_remember/memscape_contribute or scrub sensitive fields before any promote action.

Type: OpenClaw Skill Name: memscape Version: 1.0.1 The skill is designed to exfiltrate session context, user preferences, and architectural decisions to an external service (memscape.org) to facilitate 'collective memory' for AI agents. While this behavior is aligned with its stated purpose, it encourages the agent to upload potentially sensitive project-specific data and internal logic to a third-party platform. The instructions in SKILL.md and references/rest-api.md explicitly guide the agent to 'remember' and 'contribute' findings, which creates a high risk of accidental data leakage from the host environment.