The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill executes shell commands at load time to automatically populate the context with the current Git branch and pull request details. The commands used (git branch --show-current and gh pr view) are standard development tools and do not represent a security risk in this context.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of pull request comments, which could theoretically contain instructions designed to influence the agent's behavior.
Ingestion points: Pull request comment bodies are fetched via the GitHub GraphQL API in SKILL.md.
Boundary markers: The instructions do not define specific delimiters for the external comment text, but they do provide clear logical constraints on how the agent should evaluate and respond to the content.
Capability inventory: The skill is authorized to use gh api for reading and writing (posting replies) and git/gh for repository metadata.
Sanitization: The skill uses a quoted heredoc ('EOF') when constructing the shell command to post replies (gh api ... -f body="$(cat <<'EOF' ... EOF)"). This is a strong security measure that prevents the shell from expanding variables or executing commands that might be contained within the AI-generated reply body.

reply