The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill enables an agent to retrieve and process external content from GitHub (e.g., via gh issue view or gh pr view). This data is untrusted and could contain malicious instructions designed to influence the agent's behavior.\n
Ingestion points: External content is retrieved using commands that display issue and PR bodies or comments (SKILL.md, references/commands.md).\n
Boundary markers: The skill does not provide instructions to treat fetched content as data only or to ignore embedded instructions.\n
Capability inventory: The agent is instructed on how to perform write operations like merging PRs or creating releases, which could be exploited if the agent is manipulated (references/workflows.md).\n
Sanitization: No sanitization or filtering of external input is demonstrated.\n- [COMMAND_EXECUTION]: Shell Scripting and CLI Automation. The skill includes numerous examples of bash scripts that automate GitHub operations using variables derived from repository state.\n
Evidence: Examples in SKILL.md and references/workflows.md use shell features like command substitution and pipes with the gh tool.\n- [EXTERNAL_DOWNLOADS]: Interaction with a Well-Known Service. The skill facilitates cloning repositories and downloading release assets from GitHub using the official CLI tool.\n
Evidence: Usage of gh repo clone and gh release download to interact with GitHub resources (references/commands.md).

github-cli