The Agent Skills Directory

[SAFE]: The skill's logic is designed to facilitate local development workflows by reading project-specific requirements and coordinating sub-agents. It does not perform unexpected network operations or access sensitive system credentials.- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection (Category 8) due to its core design of processing external specification files to drive agent actions.
Ingestion points: The skill reads content from the specs/{feature}/ directory, including README.md, requirements.md, and various task files.
Boundary markers: Input from these files is interpolated directly into prompts for sub-agents (e.g., using placeholders like {requirements} and {task_content}) without explicit boundary delimiters or instructions to ignore embedded commands.
Capability inventory: The skill has the ability to spawn powerful sub-agents (coder and reviewer) and execute shell commands for linting, typechecking, and git version control.
Sanitization: There is no mention of sanitizing or validating the content of the specification files before they are passed to the sub-agents.
Mitigation: The risk is significantly reduced by the mandatory 'Code Review Gate' (Step 6) and 'Fix Loop' (Step 7), which utilize a separate agent to verify the integrity and safety of the implementation before any changes are committed.

implement-feature