The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several shell commands to interact with the local environment and GitHub, including git status, git log, git rev-list, and gh pr create. These are standard tools for the skill's primary purpose of development workflow automation.
[EXTERNAL_DOWNLOADS]: The skill fetches issue details and titles from GitHub using gh issue view. This interaction targets a well-known service (GitHub) to retrieve context necessary for drafting Pull Requests.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources that could be controlled by third parties, such as commit messages and GitHub issue bodies.
Ingestion points: Data is ingested in Step 2 and Step 4 via git log and gh issue view to retrieve slice references and issue content.
Boundary markers: There are no explicit delimiters or instructions to the agent to treat the fetched content as potentially untrusted data when interpolating it into the PR body template.
Capability inventory: The skill possesses the capability to execute shell commands (gh pr create) and invoke other analysis skills (/code-review, /critique).
Sanitization: There is no evidence of specific sanitization or filtering applied to the retrieved issue or commit content before it is used in prompt construction.
Mitigation: The risk is significantly reduced by mandatory human-in-the-loop (HITL) gates in Step 3 and Step 4, where the user must manually approve review findings and the final PR draft before the PR is created.

ship