Loading ToolBench

Reading the current corpus…

machvive-webmcp-ai MCP server · F · ToolBench

machvive-webmcp-ai

github.com/Mach-Five-Group/machvive-webmcp-ai

Web components and polyfill for the WebMCP (Web Machine Learning Context Protocol) API, enabling agents to discover and invoke tools registered on web pages

2026-07-28+6 tools

23/100

Definition Quality19

Protocol Readiness0

Supportability46

Spec posture

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

Static source inference · medium confidence

Current-spec patterns detected

Summary

This is a browser-based WebComponent library, not an MCP server. It provides 6 tools for managing tool registries and analytics within a web context, but lacks fundamental MCP server characteristics. Tools have minimal descriptions (many under 20 chars), no visible input schemas in the source, and no error handling guidance. The 'callTool' tool explicitly states 'Never rejects,' violating error classification patterns. No tool descriptions explain when to use them or what they return. Schemas are declared in TypeScript types but not visible in the actual tool registration code provided.

Tools (6)

callToolwritesource verified17/100

Non-standard: invokes a registered tool. Never rejects.

provideContextwrite22/100

Replaces the entire toolset. Validates before mutating.

pushToDataLayerwrite23/100

Pushes one entry to window.dataLayer in a GTM-friendly flat shape.

registerToolwrite25/100

Adds one tool, leaving the rest of the registry intact.

replaywritesource verified27/100

Re-invokes a captured call, optionally with edited params.

unregisterToolwrite30/100

Removes one tool by name; returns whether it was present.

Issues (5)

CRITICALtool ↗

No input schemas visible in source code. Tool definitions reference schema objects but actual JSON Schema definitions are not provided in the code sample. Cannot verify parameter types, constraints, or required fields.

callToolregisterToolunregisterToolprovideContextreplaypushToDataLayer

CRITICALtool-description ↗

Descriptions are too short and lack actionable context. 'Non-standard: invokes a registered tool. Never rejects.' (callTool) is 54 chars but provides no guidance on when to use it, what it returns, or error handling. Most descriptions under 100 chars.

callToolregisterToolunregisterToolprovideContextreplaypushToDataLayer

HIGHerror-classification ↗

callTool explicitly states 'Never rejects,' violating error classification and recovery patterns. Agents need to know which errors are retryable and what to do on failure. A tool that never rejects provides no error guidance.

Recommendations

Add complete JSON Schema definitions for all tool inputs. Include type, description, and constraints for every parameter. Example: registerTool's 'tool' param should define properties for name (string, required), description (string, required), inputSchema (object), and execute (function).
Expand tool descriptions to 100-200 characters. Explain WHAT the tool does, WHEN to use it, and WHAT it returns. Example: 'Invokes a registered tool by name with provided parameters. Returns the tool's result or error. Use after registerTool() to execute custom tools.'
Document output schemas for each tool. Specify what fields are returned and their types. Example: callTool returns {result: unknown, error?: string, status: 'ok'|'error'}.
Add error handling guidance. Replace 'Never rejects' with specific error cases and recovery steps. Example: 'Throws if tool not found (call registerTool first) or if parameters don't match the tool's inputSchema.'
Add parameter descriptions. The 'params' object in callTool lacks guidance on what structure is expected. Document: 'Object matching the registered tool's inputSchema. Required fields depend on the tool.'

Score history

Overall score trend

First recorded score · v2 rubric

23/100

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