The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which influences its code modification and triage logic.
Ingestion points: Reads issue references in commit messages, open pull request descriptions, and specification files located in docs/, specs/, or .scratch/ directories (Step 1). It also processes the results of findings from sub-agents (Step 2).
Boundary markers: The instructions do not define explicit delimiters or instructions for the agent to ignore potentially malicious directions embedded within the ingested specification files or PR descriptions.
Capability inventory: The skill possesses the ability to modify local code via clustered sub-agents, commit those changes using the /commit tool (Step 4), and open or refresh pull requests through /open-pr (Step 6).
Sanitization: There is no evidence of sanitization or strict schema validation for the data gathered from project history or documentation before it is passed to the sub-agents for execution.
[COMMAND_EXECUTION]: The skill invokes several external tools and sub-skills to perform automated operations.
Evidence: The workflow involves calls to /code-review, /mattpocock-skills:code-review, /browser-qa, /interview-me, /commit, /simplify, and /open-pr to process findings and apply fixes.

finish