using-glean
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides operational documentation for interacting with the Glean MCP server. The instructions focus on correct tool invocation, filter usage, and information synthesis, which is consistent with the skill's stated purpose of enterprise knowledge retrieval.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from documents, code, emails, and web pages, creating a potential surface for indirect injection.
- Ingestion points: Data retrieved via search, code_search, gmail_search, outlook_search, meeting_lookup (transcripts), and read_document (external URLs).
- Boundary markers: The skill lacks explicit technical delimiters but instructs the agent to cite every claim with a URL and prioritize official documentation over informal signals.
- Capability inventory: MCP tools for data retrieval; no evidence of arbitrary command execution or local file writing.
- Sanitization: Extensive guidelines in reference/vetting.md and reference/synthesis.md provide logical mitigations by requiring the agent to filter results for relevance, freshness, and authority, and to explicitly flag conflicting or single-source information.
Audit Metadata