litestar-mcp

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns, such as prompt injection, obfuscation, or unauthorized data exfiltration, were detected. The skill follows best practices for the Litestar framework and provides clear security guardrails for developers.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the exposure of application data and functionality to AI agents via the MCP standard. While the skill itself is a secure framework, the resulting integration creates a surface for indirect prompt injection if the developers expose endpoints that process untrusted user content without proper sanitization.
  • Ingestion points: Litestar route handlers marked with mcp_tool, mcp_resource, or mcp_prompt decorators as described in SKILL.md.
  • Boundary markers: The documentation encourages the use of explicit descriptions, structured metadata, and MCPInputRequiredResult to define clear interaction boundaries and human-in-the-loop checkpoints.
  • Capability inventory: The skill can expose any Litestar route handler as a tool, including those with side effects (POST/PUT), though it explicitly warns against exposing destructive routes without confirmation.
  • Sanitization: Relies on Litestar's robust signature and DTO validation for inputs and recommends using existing middleware and guards for access control.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 07:08 PM
Security Audit — agent-trust-hub — litestar-mcp