Worldly Wisdom

Provides calibrated decision analysis using Charlie Munger-style multiple mental models, inversion, incentive mapping, circle-of-competence checks, misjudgme...

This skill appears coherent and low-risk, but because the code source and publisher are unknown you should: (1) review the two included Python scripts before allowing execution (they only read JSON and write output but will write to paths you provide), (2) run scripts in a restricted or isolated environment if you permit local execution, (3) be aware the skill recommends using 'fresh evidence' but does not fetch it for you—if you need current facts the agent will need explicit access to a browser or data source, and (4) if you rely on the decision output for high-stakes actions, cross-check numbers and assumptions and keep sensitive credentials out of any JSON input or output paths.

Type: OpenClaw Skill Name: worldly-wisdom Version: 1.0.0 The skill bundle instructs the AI agent to execute local Python scripts (`scripts/decision_matrix.py`, `scripts/ev_scenarios.py`) via shell commands for decision analysis. While the Python scripts are well-written, perform input validation, and are purely computational without any malicious code or network access, the explicit instruction for the agent to perform local command execution (`python3 ...`) represents a risky capability. This capability, even if plausibly needed for the stated purpose, introduces a potential vector for exploitation if the agent's command construction or execution environment is vulnerable to shell injection or other local execution risks, thus classifying it as suspicious rather than benign.