The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local TypeScript scripts using the bun runtime (e.g., bun bin/plan.ts, bun bin/lib/governor.ts) to manage plan state, orchestrate leaf skills, and enforce execution envelopes.
[SAFE]: Implements a robust 'Human-in-the-loop' safety protocol that forces the agent to stop unconditionally at any 'publish' gate, ensuring no content is published or irreversible actions are taken without explicit human intervention.
[SAFE]: Enforces a mandatory approval requirement where the skill refuses to execute any plan unless its status is manually set to approved by a human, preventing the agent from self-approving or running proposed drafts.
[SAFE]: Utilizes an external governor mechanism (bin/lib/governor.ts) to check constraints such as step counts, budget caps, and domain jumps before every execution step, providing a strictly controlled execution environment.

run-plan