ComfyUI Bridge has 9 tools with mostly complete schemas and descriptions. Naming is action-oriented (generate, img2img, faceswap, etc.), which is good. Descriptions are present and moderately detailed (avg ~120 chars), meeting baseline minimums. However, several parameters lack descriptions or have vague ones. Output schemas are not documented, LLMs cannot predict what fields to expect from responses. Error handling is minimal; no recovery guidance or actionable error messages visible. Security concerns: multipart file uploads lack validation hints; no mention of permission checks or audit logging. Composition is reasonable, tools are single-purpose, but some combine related operations (faceswap_pipeline chains two steps, which could be split).

• Document output schemas for all tools. Specify response fields, types, and structure. E.g., generate returns {image_url: string, seed: int, generation_time_ms: int, model: string}.
• Add validation constraints to file upload parameters: 'Accepted formats: PNG, JPG, WebP. Max size: 10MB. Returns 400 if invalid.'
• Add error handling guidance: 'If ComfyUI is unreachable, returns 503 with retry-after header. If queue is full, returns 429. If model is missing, returns 400 with available models list.'
• Split faceswap_pipeline into two separate tools (faceswap, then img2img) so agents can compose them as needed and handle partial failures.
• Expand health and diagnostics descriptions to clarify purpose: health checks basic connectivity; diagnostics returns detailed node availability, queue depth, and system resources.