The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the GitHub CLI (gh) through a Node.js script (scripts/main.cjs) using spawnSync. The execution is configured with shell: false, which is a best practice that prevents shell injection vulnerabilities by passing arguments directly to the process.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection because its core functionality involves fetching and interpreting data from external repositories.
Ingestion points: Repository file lists, file content, and metadata are ingested via gh api and gh search (see SKILL.md).
Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted data from the agent's instructions.
Capability inventory: The skill has access to the gh CLI and the Bash and Read tools.
Sanitization: The pre-execute.cjs hook implements environment-specific safety by blocking Linux-specific constructs like /dev/stdin on Windows platforms.

github-ops