MCP server for OpenAI GPT Image generation and editing capabilities
Single tool 'generate_image' has a complete input schema with proper types, enums, and defaults. Description is adequate (194 chars) but lacks actionable guidance on when/why to use it. Parameter descriptions are present but generic, e.g., 'The text description of the desired image(s)' lacks constraints on length or content. Output schema is documented in the docstring but not formally structured in the response. No error handling guidance; no recovery paths for API failures. Tool name is clear (verb_noun), but composition could be improved: saving locally AND returning base64 conflates two concerns.
Generates an image using OpenAI's gpt-image-1 model based on a text prompt. Returns both the base64 data and saves the image locally.
Output schema not formally documented in tool registration. Docstring describes return fields, but LLM cannot parse them from schema. Missing structured output type definition.
No error handling guidance. Tool can fail with APIConnectionError, RateLimitError, APIStatusError, but no recovery hints provided to LLM. Missing pattern:recovery-guide.
Parameter 'prompt' lacks constraints. No min/max length, no guidance on content restrictions. LLM may pass empty or excessively long prompts.
Tool conflates two concerns: local file save AND base64 return. Should split into separate tools or make save_filename optional with clear semantics.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | <=2025-11-25 | v2 |
Supabase storage mode fallback logic in main.py is not exposed to LLM. Tool description does not explain where images are stored or how to retrieve them.