scholar-sidekick-mcp
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external scholarly data such as DOIs, citations, and bibliography files (BibTeX, RIS, CSL-JSON) which could contain malicious payloads. It mitigates this risk by focusing on verification and explicitly instructing the agent not to fabricate information.
- Ingestion points: User-provided scholarly identifiers and bibliography text processed as described in
SKILL.md. - Boundary markers: No explicit prompt boundary markers or delimiters are defined for the inputs.
- Capability inventory: Relies on external MCP tools (
resolveIdentifier,formatCitation,verifyCitation, etc.) to resolve metadata and format citations. - Sanitization: Sanitization and validation are expected to be handled by the external MCP server rather than the skill instructions.
Audit Metadata