Loading ToolBench

Reading the current corpus…

topaz-mcp MCP server · C · ToolBench

topaz-mcp

AI image enhancement for LLMs using Topaz Labs API. Upscales, sharpens, and improves image quality with various AI models.

2026-07-28+STDIO1 tools

69/100

Definition Quality78

Protocol Readiness45

SupportabilityNot measured-

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

Single tool 'image-enhance' has a well-structured schema with proper Zod validation and reasonable descriptions. Naming is clear (verb_noun pattern with 'enhance'). Input schema includes type constraints and ranges. However, output schema is not documented in the source, and the tool description, while adequate, could be more concise and LLM-optimized. Error handling is present but generic. No parameter enums despite having a constrained set of valid model names. The server demonstrates competent implementation but lacks some refinements expected of production-grade tools.

Tools (1)

image-enhanceread onlyauthsource verified78/100

Enhance an image using Topaz Labs AI. Upscales, sharpens, and improves quality. Accepts a local file path or URL. If the user shares an image in the conversation without providing a path, ask them for the file path or a URL to the original image.

Issues (5)

HIGHtool ↗

Output schema not documented. Tool returns success/error responses but LLMs cannot see what fields to expect or extract. No structured return type guidance.

image-enhance

HIGHconstrained-input ↗

Model parameter accepts free-form string despite having a finite set of valid options. Should be declared as an enum ('Standard V2', 'Low Resolution V2', 'CGI', 'High Fidelity V2', 'Redefine', 'Recover 3', 'Standard MAX', 'Wonder 2') to prevent LLM hallucination of invalid model names.

image-enhance

MEDIUMtool-description ↗

Size parameter is optional with no defaults documented. When omitted, behavior ('auto-upscales to 2K') is stated in description but not enforced as a schema default. LLMs may not understand the auto-scaling fallback.

image-enhance

MEDIUMtool-description ↗

Tool description mentions 'Ask the user for the path if not provided' in the image parameter description. This is a UI/agent orchestration concern, not a parameter description. Description should focus on what the parameter is, not what the agent should do if missing.

Recommendations

Document the output schema explicitly. Define what fields are returned on success (e.g., { image_url: string, width: number, height: number, processing_time_ms: number }) and on error (e.g., { error: string, error_code: string, retry_after_seconds?: number }).
Convert the model parameter to a proper enum with the eight valid options listed in the description. Replace free-form string validation with schema-level constraint: z.enum(['Standard V2', 'Low Resolution V2', 'CGI', 'High Fidelity V2', 'Redefine', 'Recover 3', 'Standard MAX', 'Wonder 2']).optional().
Add a default value for the size parameter in the schema, not just the description. E.g., { width: z.number().optional(), height: z.number().optional(), scale: z.number().optional() } with a note that if all three are omitted, the server auto-scales to 2K.
Clarify the image parameter description: remove agent orchestration hints. Change from 'Ask the user for the path if not provided' to simply 'Local file path (e.g., ~/Photos/image.jpg) or URL to the image. Local paths must be accessible to the server.'
Implement error categorization in the error handler. Return structured errors with type field: { error: string, error_code: 'INVALID_MODEL'|'FILE_NOT_FOUND'|'AUTH_FAILED'|'TIMEOUT'|'RATE_LIMIT', recoverable: boolean, hint?: string }. For INVALID_MODEL, include a hint: 'Valid models: Standard V2, Low Resolution V2, ...'
Add a dry-run or confirmation step for tools that consume API quota. Users expect destructive/costly operations (image enhancement calls an expensive external API) to have a preview or confirm before execution.

Score history

Overall score trend

↑ 23 points across a rubric change (v1 → v2)

69/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	C	69	2026-07-28+	v2
2026-03-09	F	46	-	v1