MCP server for Framehood, a platform for managing and hosting digital assets. Provides tools for file operations, asset management, and content creation.
Single tool 'files' has a verb-based name but suffers from critical definition gaps. The description (67 chars) is below the 10-1024 baseline and lacks WHAT/WHEN/WHY context. Input schema is present with types and descriptions, but the tool's purpose is unclear: it creates an upload URL, yet the description mentions 'upload.py script', mixing tool responsibility with implementation detail. No output schema is documented, forcing LLMs to guess what fields are returned. The 'create_upload' parameter is a string enum (should be declared as enum type) rather than a boolean flag. No error handling guidance. Security concern: token is returned in response, entering LLM context. Overall, this reads as a partial integration rather than a production-grade tool.
Create an upload URL and token for uploading files to Framehood. Returns upload_url and token to be used with the upload.py script.
Missing output schema documentation. Tool returns 'upload_url' and 'token' but no structured output schema is defined. LLMs cannot plan downstream calls or extract fields reliably.
Description too short (67 chars, baseline 194 avg) and lacks WHAT/WHEN/WHY context. Does not explain when to call this vs alternatives, what the token is used for, or prerequisites.
Security: 'token' parameter is returned in response, entering LLM context and logs. Short-lived tokens should never be exposed in tool output, use server-side injection or return only the upload_url.
'create_upload' parameter is a string but should be an enum with a single valid value ('create') or a boolean flag. Current design is ambiguous and invites invalid input.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
No error handling guidance. Tool description does not explain what happens on failure (invalid filename, unsupported content_type, quota exceeded) or how LLM should recover.