The Agent Skills Directory

[COMMAND_EXECUTION]: The skill mandates the execution of a shell command to validate its generated output. Evidence: The instruction to run python3 workflows/validate_tasks.py .maestro/<slug>/backend/tasks.json in the 'Emit tasks.json' section. This command executes a local repository script using the Bash tool.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its core function of reading and processing untrusted code from the repository. Ingestion points: The skill reads routing, controllers, services, and domain layers from the codebase. Boundary markers: No explicit delimiters or instructions are used to separate untrusted code content from the agent's internal logic. Capability inventory: The skill possesses Bash and Write capabilities, which could be exploited if malicious instructions embedded in the codebase are followed. Sanitization: There is no evidence of filtering or escaping logic for the content ingested from the codebase.

backend-design