deep-research
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional text providing a procedural framework for deep research. No scripts, binaries, or configuration files are present.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest external data from independent sources during the 'Gather' step, which represents an indirect prompt injection surface. However, the operating procedure includes defensive measures such as cross-checking independent sources, auditing citations, and separating evidence from inference, which serve as mitigations against external manipulation.
- Ingestion points: External sources gathered during the research process (SKILL.md, 'Gather' section).
- Boundary markers: Not explicitly defined for ingestion, but synthesis rules require clear labeling of findings vs. inferences.
- Capability inventory: No tools or command executions are defined within the skill itself.
- Sanitization: The skill mandates a citation audit to verify sources are locatable and authentic before delivery.
Audit Metadata