plan
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to generate implementation plans and execute code modifications, creating an attack surface for indirect instructions.
- Ingestion points: Untrusted content is ingested from GitHub Action logs (references/mode-ci.md), arbitrary web URLs via web_fetch (references/mode-cro.md), and remote GitHub repositories via repomix (references/engine-research.md).
- Boundary markers: The instructions do not specify the use of boundary markers or protective delimiters when processing the fetched logs or website content.
- Capability inventory: The skill has significant capabilities, including file system modification, Git operations (git-manager), and the power to spawn sub-agents (spawn_agent) for code implementation (references/mode-execute.md).
- Sanitization: There is no explicit requirement to sanitize or filter the content retrieved from external sources before it is used by the planning or implementation agents.
- [COMMAND_EXECUTION]: The skill executes command-line utilities to perform research and manage the repository.
- Evidence: references/engine-research.md utilizes 'gemini' for automated search and 'repomix' for repository summarization. references/mode-execute.md uses 'git-manager' to handle repository operations.
- [EXTERNAL_DOWNLOADS]: The skill fetches content from remote servers to provide context for its tasks.
- Evidence: references/mode-ci.md downloads CI logs; references/mode-cro.md fetches web content for analysis; references/engine-research.md clones or downloads repository data using 'repomix --remote'.
Audit Metadata