gsp-project-research
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection surface identified. The skill ingests untrusted data from external websites and project files which could influence the behavior of the researcher agent.
- Ingestion points: Data is read from project-level files such as
BRIEF.mdand{PROJECT_PATH}/references/, and external content is retrieved using theWebSearchandWebFetchtools. - Boundary markers: The skill instructions do not define clear delimiters or provide instructions to the agent to disregard potential commands embedded within the researched content.
- Capability inventory: The skill possesses the capability to write multiple files to the local filesystem and has access to a wide range of tools including
BashandWritevia the spawned researcher agent. - Sanitization: There is no evidence of sanitization or validation of the fetched external content before it is processed or used to generate local project files.
- [DATA_EXFILTRATION]: The skill performs network operations to external, non-whitelisted domains.
- Network operations: Uses the
WebSearchandWebFetchtools to interact with external websites to retrieve research data and pre-fetch references mentioned in the project brief.
Audit Metadata