Rationality

Apply Critical Fallibilism to make decisions by binary testing ideas for decisive flaws, managing complexity, embracing criticism, and avoiding overreach.

This skill is a collection of frameworks and templates for binary/critical-fallibilist thinking and appears internally consistent. Before installing, check how your agent runtime implements 'memory/' and whether it allows git and shell execution: the SKILL.md suggests writing refutations to memory/ and sometimes using git/reset or exec. If you prefer to limit filesystem or shell access, run the skill in a restricted sandbox or disable any automated writing and treat the files as read-only guidance. No credentials or network installs are required, so the main operational consideration is whether you want the agent to persist changes or run local commands as the documentation suggests.

Type: OpenClaw Skill Name: rationality Version: 0.1.0 The skill bundle provides a philosophical framework and operational guidelines for an AI agent's decision-making and error correction. While there is no clear evidence of malicious intent such as data exfiltration or unauthorized remote control, the `patterns/overreach.md` file contains an instruction for the agent to "Use `exec` to probe the environment" when a tool fails. This grants the agent broad command execution capabilities, which, despite being framed for debugging and understanding tool behavior, represents a significant security risk due to its potential for misuse if not properly sandboxed.