The Agent Skills Directory

[COMMAND_EXECUTION]: The skill guides the agent to use a CLI tool named agentit for running verification probes and managing compliance receipts. This command execution is scoped to the project environment and serves the primary function of the skill to ensure work quality.
[INDIRECT_PROMPT_INJECTION]: The skill handles potentially untrusted data from project files and task descriptions. 1. Ingestion points: Task input in agentit verify "task" and config files like probes/catalog.yaml. 2. Boundary markers: Absent. 3. Capability inventory: Execution of local project scripts. 4. Sanitization: Not explicitly present, but mitigated by the 'anti-greenwash' rules and the mandatory requirement for observed evidence before claiming completion.

verification-gauntlet