feature-flags
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The flag-cleanup workflow ingests local source code files using the Read tool to perform inlining of feature flag values. Malicious content within these project files could potentially influence agent behavior during the cleanup process.\n
- Ingestion points: The instructions in references/flag-cleanup.md direct the agent to use the Read tool to load project source files for analysis and modification.\n
- Boundary markers: The workflow does not specify the use of delimiters or explicit "ignore embedded instructions" warnings when processing the content of these files.\n
- Capability inventory: The skill is granted access to the Bash tool (via the gb-call helper), as well as the Edit, Grep, and Read tools, which creates an exploitable surface if the agent obeys instructions injected through file content.\n
- Sanitization: No methods for sanitizing, escaping, or validating the external file content before agent processing are described in the instructions.
Audit Metadata