The Agent Skills Directory

[SAFE]: The skill is entirely instructional, establishing the 'Prove It Works' principle which encourages agents to verify their own output through direct observation rather than relying on self-reports or indirect evidence.
[NO_CODE]: The skill does not bundle any executable scripts, binaries, or configuration files. It is a pure markdown instruction file.
[COMMAND_EXECUTION]: While the skill suggests that the agent should 'Script the check when you can,' it does not include any pre-defined commands or scripts. The recommendation is a standard software engineering practice for ensuring deterministic verification of work performed in the local environment.

principle-prove-it-works