skill-copilot-provider
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill integrates GitHub Copilot by executing the
copilotcommand-line utility in programmatic mode. It specifically utilizescopilot -pwith the--no-ask-userflag to allow the agent to obtain responses without interactive prompts, which is standard for non-interactive provider integrations. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates an interface where user-provided prompts are interpolated into a shell command execution environment. While this introduces an attack surface common to provider wrappers, it is intended functionality.
- Ingestion points: The prompt content passed to the
copilot -pcommand as documented inSKILL.md. - Boundary markers: No explicit delimiters or boundary markers are defined in the command template to separate user input from the command context.
- Capability inventory: Shell execution of the
copilotbinary with access to local environment variables for authentication. - Sanitization: The skill does not explicitly provide or document sanitization for the input prompt before command interpolation.
Audit Metadata