Srs

Automation skill for Srs.

This skill appears to implement a local Security Research System and is coherent with that purpose, but it will: (1) scan folders in your home directory (~/ai-security/research, ~/.openclaw/workspace, etc.) and read .md/.json/.yaml files; (2) write persistent files under ~/.openclaw/workspace/srs (capabilities.json, role files, results); and (3) execute local Python scripts via subprocess (e.g., tasks/run_*.py) if present. Before installing: review the included Python files (srs.py, self_governor.py, parallel_executor.py, coordination.py) to confirm behavior; ensure there are no unexpected network calls or references to external endpoints; run the skill in a restricted/sandboxed environment or with limited filesystem permissions if possible; check for any local scripts in tasks/ that could be executed; and prefer skills with known provenance. If you want higher assurance, ask the author for a provenance statement or sign-off, or request a minimal mode that only performs read-only discovery (no subprocess execution or writes).

Type: OpenClaw Skill Name: srs Version: 1.0.0 The bundle implements a 'Security Research System' (SRS) for automating task evaluation and role management. It contains high-risk behaviors including broad filesystem scanning in `srs.py` and `self_governor.py` (targeting paths like `~/ai-security/research` and `~/.openclaw/workspace`) to discover and create new role configurations. Furthermore, `parallel_executor.py` uses `subprocess.Popen` to run arbitrary Python scripts from a `tasks/` directory. While these functions align with the stated purpose of a security research tool, the combination of automated filesystem discovery and local code execution represents a significant attack surface.