Async MCP server with Minimax API integration for image generation and text-to-speech
Server demonstrates solid definition quality with comprehensive parameter schemas and descriptions for all three tools. Tool naming follows action-verb conventions (submit_*, task_barrier). Descriptions are detailed and contextual, explaining the async batch-processing pattern. However, there are notable gaps: no output schemas are documented, error handling lacks recovery guidance, and the tool set has a minor composition concern (task_barrier is a coordination primitive rather than domain-specific).
Generate images asynchronously. RECOMMENDED: Submit multiple tasks in batch to saturate rate limits, then call task_barrier once to wait for all completions. Returns task ID only - actual files available after task_barrier.
Convert text to speech asynchronously. RECOMMENDED: Submit multiple tasks in batch to saturate rate limits, then call task_barrier once to wait for all completions. Returns task ID only - actual files available after task_barrier.
Wait for ALL submitted tasks to complete and retrieve results. Essential for batch processing - submit multiple tasks first, then call task_barrier once to collect all results efficiently. Clears completed tasks.
No output schemas documented for any tool. Callers cannot predict response structure, making downstream tool chaining difficult and forcing LLMs to infer field names from text responses.
Error responses are formatted as plain text without structured recovery guidance. LLMs receive '❌ Failed to submit image generation task: <error>' but no hint about what to do next (retry? check parameters? contact admin?).
Parameter interdependencies not fully documented. For submit_image_generation, 'customSize' and 'aspectRatio' are mutually exclusive (aspectRatio takes precedence), and 'style' cannot be combined with customSize or subjectReference. These constraints are mentioned in descriptions but not in a machine-readable format (oneOf/not/allOf in schema).
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 63 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 45 | - | v1 |
submit_speech_generation has 18 optional parameters, many with unclear interaction semantics. For example, languageBoost has value 'auto' as default but enum includes 26 language options, unclear when and how the auto-detection works relative to explicit selections.
task_barrier is a coordination primitive, not a domain tool. Its name does not convey 'wait for async tasks' naturally, an LLM might not recognize it as the completion-blocking mechanism without explicit context. Consider renaming to 'wait_for_tasks' or 'get_task_results'.