The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the target repository's .tailrocks/pr.md file and executes the tasks listed in its ## Before merge section.
Ingestion points: The agent is instructed to read .tailrocks/pr.md (Step 4) to determine the pre-merge worklist.
Boundary markers: The skill includes a "Red flags — STOP" section that explicitly states PR content (body, comments, reviews) is evidence, not instructions. It also instructs the agent to "Classify each item before execution" (Step 4).
Capability inventory: The agent has access to the gh CLI (view, diff, edit, merge), file system write operations, and the ability to commit/push via git. It also executes specific TypeScript entrypoints (merge-preflight and merge-pr).
Sanitization: The skill requires explicit user authorization for high-blast-radius actions or irreversible effects like deployment or data mutation.
[COMMAND_EXECUTION]: The skill performs extensive operations using the gh (GitHub) CLI, including merging pull requests and editing metadata. It also performs git commit and git push operations as part of the repository-defined worklist.
[DYNAMIC_EXECUTION]: The skill executes external TypeScript scripts (merge-preflight and merge-pr) located in a scripts directory determined at runtime relative to the SKILL.md file path. It includes a security check to ensure these entrypoints are regular files and not symbolic links before execution.

tailrocks-merge-pr