feature-agents
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The code-explorer and code-architect agents are designed to ingest and process information from local source code files to provide architectural insights. This functionality presents a standard surface for indirect prompt injection if the files being analyzed contain adversarial content designed to influence the agent's reasoning.
- Ingestion points: Local source code files and directory structures accessed during discovery and flow tracing as defined in references/code-explorer.md and references/code-architect.md.
- Boundary markers: The system prompts do not currently implement specific delimiters or 'ignore' instructions for the content of the files being analyzed.
- Capability inventory: The agents utilize the provided context to map dependencies, trace data flow, and generate implementation blueprints.
- Sanitization: There is no explicit sanitization of codebase content within the provided prompts.
Audit Metadata