The Agent Skills Directory

[SAFE]: The skill instructions define a procedural 'Verification Gate' that requires the agent to run commands and verify output before claiming completion. This is a safety-enhancing design pattern.- [SAFE]: File system access is restricted to checking metadata (existence and line count) of result files in a specific local application directory (~/.claude-octopus/results/).- [SAFE]: The provided command examples, such as npm test, ls, and wc, are standard development tools used for their intended purpose of verification.

skill-verification-gate