The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub Pull Requests (titles, bodies, and diffs) and passes it to delegated AI models for review.
Ingestion points: Untrusted content enters the agent's context via gh pr view and gh pr diff commands as described in the review workflow.
Boundary markers: The review brief sent to delegated models in the Convene section lacks explicit delimiters or instructions to ignore embedded commands within the PR content, which could lead to models obeying instructions hidden in the code or PR description.
Capability inventory: The agent has the authority to merge code (gh pr merge), modify files for fixes, and delegate tasks to other models.
Sanitization: No sanitization or escaping of the PR content is performed before it is interpolated into the review prompts provided to the council members.
[COMMAND_EXECUTION]: The skill relies on executing shell commands to manage the Git repository and GitHub Pull Requests.
Evidence: The instructions utilize gh pr view, gh pr diff, gh pr merge, git rev-parse, and git status. These commands are used to interact with the host environment and external GitHub services.

council