The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the git and gh CLI tools to manage repository state, including the creation of git worktrees in the parent directory and performing force-pushes to the remote origin. These actions are within the scope of a merge queue manager but involve significant privileges. It also uses a custom tool named 'bd' for managing the lifecycle of project issues.
[EXTERNAL_DOWNLOADS]: The skill fetches data from GitHub, which is a well-known service. This includes pull request metadata, commit history, and CI run status/logs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted content from external contributors. (1) Ingestion points: The skill reads PR bodies (Step 1/5), commit messages (Step 4), and CI failure logs (Step 7). (2) Boundary markers: There are no delimiters or instructions used to prevent the agent from following commands embedded in these data sources. (3) Capability inventory: The skill can merge code, force-push to branches, and spawn sub-agents with specific roles. (4) Sanitization: No sanitization or validation is applied to the external data before it is used to drive the agent's logic or decision-making.

merge-queue