The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute verification commands found within the project's local files, such as CLAUDE.md or existing scripts. This is its intended behavior for providing evidence-backed proof of implementation.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it relies on instructions parsed from external files to determine which commands to execute.
Ingestion points: The agent is instructed to read CLAUDE.md and other local scripts to discover execution logic.
Boundary markers: There are no explicit markers or validation steps defined to separate instructions from data within the ingested files.
Capability inventory: The skill utilizes Bash for execution and Write to generate .maestro/ artifacts.
Sanitization: No sanitization or verification of the discovered commands is performed before they are passed to the shell.

verify