The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various CLI tools including git for worktree management and gh for pull request interactions, status checks, and API calls.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its autonomous processing of external, untrusted content: 1. Ingestion points: The agent reads untrusted data from pull request diffs (gh pr diff), file contents within PRs, and CI failure logs. 2. Boundary markers: There are no instructions to use delimiters or ignore instructions embedded within the ingested PR content or logs. 3. Capability inventory: The skill has broad capabilities across all scripts including creating branches, pushing code, posting comments via GitHub API, and merging pull requests. 4. Sanitization: The skill lacks explicit steps to sanitize or validate external content before it influences the agent's logic or subsequent sub-agent tasks.

forge-loop