The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests pull request descriptions, metadata, and code diffs from external contributors. These inputs are used to generate reports and propose code changes that the agent then commits and merges. A malicious contributor could attempt to embed instructions within a PR to manipulate the agent's actions during the fix application or merge phases.
Ingestion points: Ingests untrusted data via gh pr view and gh pr diff in Step 3.
Boundary markers: A mandatory human-in-the-loop (HITL) checkpoint is defined in Step 6, requiring user approval before the agent proceeds to modify code.
Capability inventory: The skill has capabilities to write files, execute git commit, git push, and gh pr merge in Steps 7 and 8.
Sanitization: No automated sanitization or filtering of the ingested PR content is performed before processing.
[COMMAND_EXECUTION]: The skill executes several shell commands to manage the development environment, including npm install, git remote, and gh pr edit. While standard for development workflows, these commands modify the local environment and the remote repository state.
[DYNAMIC_EXECUTION]: The skill uses node -p in Step 2 to dynamically evaluate JavaScript code for the purpose of parsing package.json and calculating version increments.

review-prs-cx