grill-summary
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted content from 'grills' and project 'maps' (Markdown files or GitHub issues). This provides an attack surface where malicious instructions embedded in those documents could attempt to influence the agent's behavior during summarization.
- Ingestion points: The skill reads user-supplied artifacts, Markdown files in the '.grills/' directory, and GitHub issue content (referenced in SKILL.md and WAYFINDER.md).
- Boundary markers: No explicit delimiters are specified to isolate the processed text from the agent's system instructions.
- Capability inventory: The agent is granted read access to the local file system and is instructed to use 'gh' read commands to fetch GitHub issue metadata. The instructions explicitly state the skill is 'read-only' and prohibit any write operations or implementation steps.
- Sanitization: There are no explicit instructions for sanitizing or filtering input text before processing.
- [COMMAND_EXECUTION]: The skill directs the agent to use the 'gh' command-line interface to retrieve metadata from GitHub issues. This is a legitimate tool invocation for the skill's purpose of providing project-wide context and involves a well-known, trusted service.
Audit Metadata