The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to use the copilot CLI with the --allow-all flag. This configuration enables the agent to perform actions such as code modifications and pull request management without human-in-the-loop confirmation.
[PROMPT_INJECTION]: The skill demonstrates patterns where dynamic content is placed into a CLI prompt (e.g., copilot --prompt "..."). This represents an ingestion point for indirect prompt injection. If the agent populates these prompts with untrusted data from the repository or external sources without sanitization or boundary markers, the copilot tool could be coerced into performing unintended actions.

copilot-cli