The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from GitHub PR reviews and comments. In SKILL.md, Phase 3 and 4 retrieve external content using bun gh-tool pr threads and bun gh-tool pr issue-comments. The agent is instructed to "auto-apply" suggestions for bugs, style, or security fixes without user confirmation, enabling a scenario where malicious instructions in a comment could lead to unauthorized code changes being committed and pushed.
Ingestion points: PR review threads and issue comments in SKILL.md.
Boundary markers: None; the agent is told to directly parse and act on suggestions.
Capability inventory: git commit, git push, and bun gh-tool pr reply-and-resolve across SKILL.md and references/push-branch-sync.md.
Sanitization: None.
[COMMAND_EXECUTION]: The workflow involves extensive use of shell commands. It uses standard git for repository management and custom CLI tools like bun gh-tool for GitHub operations and but (GitButler) for branch synchronization as described in references/push-branch-sync.md.

git-workflow