The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface by ingesting and acting upon untrusted data from PR review comments.
Ingestion points: Fetches external content via gh api for both inline and general PR comments in SKILL.md.
Boundary markers: Instructions lack explicit delimiters or boundary markers to isolate external comment content from the agent's internal reasoning logic.
Capability inventory: The skill has the capability to modify the codebase, execute the /commit skill, and perform git push operations.
Sanitization: No sanitization or filtering of the comment body is specified before the agent analyzes the feedback and generates code diffs.
Mitigation: The workflow includes a mandatory interactive approval step using the AskUserQuestion tool, ensuring a human reviews all proposed fixes before they are applied or pushed to the repository.
[COMMAND_EXECUTION]: The skill executes shell commands to interact with the environment and GitHub services.
GitHub API: Uses gh api to retrieve review data and post replies to PR threads.
Git Operations: Executes git push to synchronize local changes with the remote repository.

resolve-pr-reviews