Safe Evolver

A safe AI agent evolution engine that analyzes runtime history to identify improvements and applies protocol-constrained evolution with comprehensive safety...

This package appears to do what it says: record interactions, analyze local history, and generate suggestions. Before installing or using it: - Expect it to create ./logs and ./data directories and to persist interaction data; do not record sensitive secrets or file contents into the evolver. Change logPath/historyFile to a safe location and set restrictive file permissions if needed. - Note that SKILL.md mentions LLM integration and export features that are not implemented in the provided index.js; if you need those capabilities, verify the upstream source or implement them yourself. - Because it writes files, run it in a controlled environment (or container) first and review the created files to confirm no unexpected behavior. If you plan to use it in production, inspect the repository (homepage/repo) and confirm provenance before trusting logs/history data.

Type: OpenClaw Skill Name: safe-evolver Version: 1.1.1 The 'safe-evolver' skill is a telemetry and logging utility designed to record and analyze AI agent interactions, such as tool calls, errors, and efficiency metrics. While the documentation (SKILL.md) describes an elaborate 'evolution engine' with LLM-based analysis, the actual implementation in index.js is a simple local logger that stores history in JSON files and provides randomized improvement suggestions. There is no evidence of data exfiltration, unauthorized execution, or prompt injection; sensitive file paths like /etc/passwd are mentioned only as illustrative examples in the documentation.