find-simplifications
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted content from GitHub issues and the repository's own files during its survey phase. This constitutes a surface for indirect prompt injection if malicious instructions are embedded in issue comments or source code.\n
- Ingestion points: Reads from a GitHub issue 'ledger' and scans the entire repository codebase and documentation.\n
- Boundary markers: The instructions explicitly command the agent to treat issue text, PR text, and code comments as untrusted and to ignore any instructions found within them.\n
- Capability inventory: The skill can execute Git commands, interact with GitHub issues/PRs, and perform file system operations.\n
- Sanitization: Mitigation relies on explicit agent instructions to ignore injected content and a mandatory human review step before any code is actually deleted.\n- [COMMAND_EXECUTION]: The skill performs numerous shell operations using standard tools including
git,gh,npm, andripgrep. These operations are used for repository state management (worktrees), testing, and reporting, all within the context of the project's maintenance needs.
Audit Metadata