api-server-mcp
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes an API server and MCP integration that ingests external data (multipart files and URLs) for extraction, which creates a potential surface for indirect prompt injection where adversarial instructions in the processed data could influence the agent's behavior.
- Ingestion points: The documentation specifies several endpoints and tools for data ingestion, including
POST /extract,POST /detect, and theextractandextract_batchMCP tools. - Boundary markers: The skill does not explicitly define delimiters or instructions for the agent to ignore embedded commands within the extracted content.
- Capability inventory: The skill indicates that the server performs filesystem writes for caching (
crates/xberg/src/cache/) and network operations to fetch content from URLs. - Sanitization: While the skill includes a 'Critical Rule' to validate uploads (MIME type, size, and magic bytes), it focuses on file integrity rather than sanitizing extracted text for prompt injection patterns.
Audit Metadata