The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill follows developer best practices by requesting user confirmation before submitting data to a remote repository.
[COMMAND_EXECUTION]: Uses legitimate development tools (git and gh) to gather information about the current workspace and create pull requests. The commands are scoped to the project's context.
[DATA_EXFILTRATION]: All network activity is directed to the project's official GitHub repository via the gh tool. No sensitive files from outside the repository (such as SSH keys or environment secrets) are accessed or transmitted.
[PROMPT_INJECTION]: The instructions do not contain attempts to override system behavior or bypass safety filters.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from git log and git diff (Step 3).
Ingestion points: SKILL.md Step 3 reads content from the local git repository.
Boundary markers: None explicitly defined in the bash calls, but content is used for drafting.
Capability inventory: The skill can create pull requests via gh pr create (Step 5).
Sanitization: Risk is mitigated by Step 4, which requires the user to review and manually confirm the drafted content before the PR is created.

pr-to-develop