Loading ToolBench

Reading the current corpus…

modelmux MCP server · F · ToolBench

modelmux

MCP server that lets Claude Code, Codex, and Perplexity call each other as tools. Supports text, code, images, and PDFs passed by file path.

2026-07-28+STDIO4 tools

47/100

Definition Quality51

Protocol Readiness50

Supportability38

Spec posture

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

Static source inference · medium confidence

Current-spec patterns detected

Summary

Four tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. All tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization). Input schemas present but parameters lack detailed constraints and format guidance. No output schemas documented. Error handling is minimal, no recovery guidance or actionable error messages. Tool names are reasonably clear (ask_*, broker) but descriptions are terse. The 'filePath' parameter appears across all tools but lacks validation rules, format constraints, or guidance on supported file types.

Tools (4)

ask_clauderead onlyauthsource verified52/100

query Anthropic's Claude API

ask_codexread onlyauthsource verified52/100

query OpenAI's GPT API

ask_perplexityread onlyauthsource verified52/100

query Perplexity's web-grounded search API

brokerread onlyauthsource verified47/100

query multiple AIs in parallel, then synthesize results

Issues (5)

HIGHtool-description ↗

Tool descriptions are under 50 characters and lack context for LLM selection. 'query Anthropic's Claude API' does not explain WHEN to use ask_claude vs broker, what it returns, or prerequisites (API key required). Baseline for A+ tools is 50-200 chars with explicit WHAT/WHEN/RETURN guidance.

ask_claudeask_codexask_perplexitybroker

HIGHconstrained-input ↗

filePath parameter lacks validation rules, format constraints, and supported file type documentation. Code shows support for text, images, PDFs with different handling per provider, but the parameter description does not mention these constraints. LLMs will pass unsupported formats without guidance.

ask_claudeask_codexask_perplexitybroker

HIGHtool ↗

No output schemas documented. LLMs cannot plan downstream tool calls or extract structured data without knowing what fields to expect. Code shows API responses are returned but schema is not declared in tool definitions.

ask_claude

Recommendations

Expand tool descriptions to 80-150 characters. Example: 'ask_claude: Query Claude via Anthropic API. Use when you need reasoning, code analysis, or multi-turn context. Returns the model's text response. Requires ANTHROPIC_API_KEY env var.'
Document filePath parameter constraints: 'Optional path to a file (text, code, image, or PDF). Supported: .txt, .js, .py, .png, .jpg, .pdf. Max 20 MB. Text files inlined; images/PDFs sent as base64. Perplexity accepts text/code only.'
Add output schema to each tool definition. Example: '{ type: "object", properties: { response: { type: "string", description: "The model's text response" }, usage: { type: "object", properties: { input_tokens: { type: "number" }, output_tokens: { type: "number" } } } }'
Enhance error messages with recovery guidance. Example: 'API key not found. Set ANTHROPIC_API_KEY=sk-... before launching the MCP server.' or 'File not found at /path/to/file. Check the path and try again.'
Clarify systemPrompt: 'Optional system-level instruction that shapes the model's behavior (e.g., "You are a Python expert"). Sent to the API before the main prompt. Omit for default behavior.'

Score history

Overall score trend

First recorded score · v2 rubric

47/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	47	2026-07-28+	v2