The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard search and navigation commands (e.g., grep, directory listings) to identify project documentation, architectural decision records, and downstream tasks for compatibility analysis.
[INDIRECT_PROMPT_INJECTION]: The skill processes external artifact files such as specifications, plans, and source code as primary input for validation. (1) Ingestion points: conversation context and specified artifact files like tasks.md or plan.md. (2) Boundary markers: implicit constraints are provided by defined validation lenses and output templates rather than explicit delimiters. (3) Capability inventory: utilizes the task tool to launch parallel agents and includes the ability to modify project files to apply fixes. (4) Sanitization: relies on structured analysis logic and specific reviewer prompts to mitigate the risk of embedded instructions.

validate-v2