The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard command-line tools such as git and gh (GitHub CLI) to manage the development workflow. It includes explicit instructions to stop and ask for user confirmation before performing sensitive actions like committing changes or pushing to a remote branch.
[DATA_EXPOSURE]: To function, the skill must read repository status, PR branch details, and CI logs. This data ingestion is restricted to the context of the PR being managed and is consistent with the skill's primary purpose.
[PROMPT_INJECTION]: The skill interacts with external data sources such as pull request descriptions and CI log outputs, which constitutes an attack surface for indirect prompt injection (Category 8).
Ingestion points: GitHub CLI log outputs and PR branch contents.
Boundary markers: No explicit delimiter markers are defined in the prompt for log data.
Capability inventory: File system modification, git commits, and network pushes via git/gh.
Sanitization: The risk is mitigated by the instruction to require human confirmation before any destructive or public-facing action (pushing/committing).

pr-ready