The Agent Skills Directory

[COMMAND_EXECUTION]: Uses the GitHub CLI (gh) and a local Node.js script to aggregate information. Shell commands are executed with controlled arguments via execFileSync, preventing arbitrary command injection. The use of gh, node, and bun is explicitly restricted via the allowed-tools frontmatter.
[DATA_EXFILTRATION]: Accesses GitHub repository metadata (issues, PRs, project boards). The skill does not access sensitive local files (like SSH keys) or exfiltrate data to third-party servers; all network communication is handled through the official GitHub CLI tool.
[PROMPT_INJECTION]: The skill processes untrusted input from GitHub, including issue titles and label names, which constitutes an indirect prompt injection surface. This risk is mitigated by explicit instructions requiring the agent to obtain user confirmation before performing any write operations, such as commenting on issues or editing labels.

gh-inbox