The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to interact with Git and the GitHub CLI. It executes commands like 'git status', 'git push', and 'gh pr create' to manage the code lifecycle. Additionally, it runs repository-specific 'lint' and 'test' scripts to ensure code quality before pushing changes.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources such as ticket titles from Linear or GitHub and local commit messages to populate pull request drafts. The risk is mitigated by a mandatory evidence chain: the skill must display the complete draft and wait for explicit user approval before final submission, ensuring a human-in-the-loop verification of all aggregated data. (1) Ingestion points: Fetches ticket metadata from external issue tracker URLs and local Git history. (2) Boundary markers: Mandatory draft display and explicit approval checkpoints are defined in the doctrine. (3) Capability inventory: Branch management, network pushing, and PR creation via tools. (4) Sanitization: Relies on user verification of the interpolated content before the publish command is executed.

publish