The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent on how to use the copilot CLI. It highlights that the --yolo flag enables arbitrary command execution and workspace mutation, advising that it should only be used with explicit user authorization.
[DATA_EXFILTRATION]: The skill sends local project excerpts to GitHub Copilot's external servers. It includes mandatory steps for the agent to obtain user consent before any project files or snippets are shared externally.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it reads and processes local file content which could contain malicious instructions. * Ingestion points: File excerpts are read via sed in the provided examples in SKILL.md. * Boundary markers: The skill suggests using static instruction text to delimit reviewed content. * Capability inventory: The agent is granted file read access via sed and shell execution via the copilot command. * Sanitization: The instructions explicitly advise against interpolating untrusted text into shell commands and recommend using quoted variables.

ask-copilot