Skip to content

Security audit findings — tool description injection + missing output sanitization #273

@manja316

Description

@manja316

Hi tadata-org team,

I recently ran a security audit on fastapi-mcp as part of research on MCP server framework security posture.

Found a couple of items worth flagging:

1. Tool description injection risk
fastapi-mcp auto-generates tool descriptions from FastAPI endpoint docstrings and OpenAPI specs without validating them against adversarial patterns. Any FastAPI app with user-controlled endpoint descriptions could generate poisoned tool descriptions that redirect LLM behaviour.

2. Missing output sanitization
API response content forwarded to the model context isn't scanned for embedded injection patterns. Framework-level sanitization here would protect every FastAPI app exposed as an MCP server — zero effort for downstream developers.

Both are in a full audit report — 8-page PDF with CVSS ratings, EU AI Act mapping, and remediation steps — for $29 at luciferforge.github.io/mcp-security-audit/

Demo: https://luciferforge.github.io/mcp-audit-reports/

— Lucifer / LuciferForge Security

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions