Workspace Analyzer

Analyzes OpenClaw workspace structure and content to identify maintenance needs, bloat, duplicates, and organization issues. Outputs a JSON report for the ag...

What to consider before installing or running: - Inspect run.sh before executing: the provided run.sh contains what looks like an embedded Python snippet and appears malformed; do not run the script blindly. Prefer running analyzer.py directly (python3 scripts/analyzer.py ...) if you trust the code. - Limit the scan root: by default it targets ~/.openclaw/workspace, which is appropriate. Avoid passing --root=/ or other system-wide paths unless you explicitly intend to scan them—the analyzer will read any files it is pointed at. - Review output before allowing automated fixes: the analyzer only reports issues and recommendations, but SKILL.md includes guidance for an agent to act. Only permit an automated agent to modify files (consolidate, delete, edit) after you have reviewed the JSON report. - Confirm no external transmission: analyzer.py reads local files and writes a JSON report (e.g., /tmp/workspace-analysis.json) and does not include network calls. If you later modify the skill to add 'fix' actions, review any new code for upload/endpoints. - If you have strict security policies, run the script in an isolated environment (container or VM) first and inspect results and any code changes before allowing the agent to act on your real workspace.

Type: OpenClaw Skill Name: workspace-analyzer Version: 1.1.0 The OpenClaw AgentSkills bundle is classified as benign. The `SKILL.md` clearly outlines the purpose of analyzing the workspace and providing recommendations for an AI agent to act upon. Crucially, the `scripts/analyzer.py` Python code is strictly read-only, performing file system traversal and content analysis without any write operations, network calls, or external command execution. This aligns with the 'Safety & Security' claims in `SKILL.md` ('Read-Only', 'No Secrets', 'Safe Output'). While `SKILL.md` includes 'Fix Instructions for Agents' involving `git add -A` and `git commit`, these are instructions for the AI agent to perform *after* reviewing the analysis, not actions executed by the provided script itself. There is no evidence of intentional harmful behavior such as data exfiltration, backdoors, or unauthorized remote control.