MCP server for image analysis using Vision Language Models (VLM). Provides tools to describe images from files, clipboard, or base64-encoded data.
Three tools with minimal descriptions (all under 100 chars) and no visible input parameter descriptions. Schemas are present but lack parameter-level documentation. Tool names follow verb_noun pattern (describe_*) but descriptions are generic and do not explain WHEN to use each variant or what distinguishes them. No output schema documentation visible. Error handling not evident in tool definitions.
Describe a base64-encoded image using Vision Language Model
Describe an image from the system clipboard using Vision Language Model
Describe an image from a file path using Vision Language Model
Parameter descriptions missing. 'image_path' and 'base64_image' parameters lack descriptions explaining format, constraints, or expected values. LLMs cannot infer whether image_path accepts relative or absolute paths, or what image formats are supported.
Tool descriptions are generic and under 100 characters. They do not explain WHEN to use each variant (file vs clipboard vs base64), what the output contains, or what VLM is being used. Descriptions should be 50-200 chars with clear intent.
No output schema documented. LLMs do not know what fields the response contains, what format the description is in, or how to chain results to downstream tools. Output structure must be explicitly documented.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | <=2025-11-25 | v2 |
No error handling guidance. If an image file does not exist, clipboard is empty, or base64 is invalid, the LLM receives no recovery hint. Error responses should guide the agent to retry or use an alternative tool.
Three nearly identical tools with no clear differentiation. Descriptions do not explain why an agent should choose file over clipboard over base64. This forces the LLM to guess or try all three, wasting tokens and latency.