interview
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project-level files which could be manipulated by an attacker to influence the agent's behavior during the interview.
- Ingestion points: The skill reads
spec/vision.md,spec/glossary.md, and general project source code to ground its questions. - Boundary markers: There are no explicit instructions or delimiters used to separate external file content from system instructions or to warn the agent about potential embedded prompts.
- Capability inventory: The skill possesses file-write permissions to the
spec/and.scratch/directories and the ability to execute shell commands provided by theexploretool. - Sanitization: No sanitization, validation, or filtering of the content read from files is specified.
- [DYNAMIC_EXECUTION]: The skill relies on a command provided by the
exploretool at runtime to render visualizations. - Evidence:
references/tree.mdinstructs the agent to obtain a render command from the 'explore' tool and re-run it repeatedly after updating state files.
Audit Metadata