Loading ToolBench
Developer utility workspace with 40+ tools for code formatting, data transformation, validation, and conversion
devdrivr is a Tauri desktop application, not an MCP server. The codebase contains 33 tools registered in src/app/tool-registry.ts, but these are React components for a local UI, not MCP tool definitions. No tool has an explicit input schema, parameter descriptions, or output schema visible in the source. Tool descriptions exist but are UI-focused summaries (e.g., 'Format and beautify code') rather than LLM-optimized docstrings. The package.json lists rmcp (MCP Rust library) as a dependency, but no MCP server implementation is evident in the provided source. Without visible tool schemas, parameter types, or structured definitions, the definition quality is severely limited.
Make HTTP requests with history, auth, headers, body, response preview, and export
Encode/decode Base64 with URL-safe, line wrap, image preview, data URIs
Convert between 12 cases with detection, word splitting, and chaining
Format and beautify code (JS, TS, JSON, CSS, HTML, YAML, SQL, GraphQL)
Convert hex/rgb/hsl/hsb/lab/lch/oklch with named colors, scales, harmony, and history
Calculate specificity with segmented bars, component breakdown, winner detection, and !important
Convert CSS rules to Tailwind classes
No input schemas visible for any tool. All 33 tools lack JSON Schema definitions with parameter types, constraints, and descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 35 | 2026-07-28+ | v2 |
Check CSS against the specification with clickable problems, a specificity ranking, and formatting
Sortable, filterable table beside the source, conversion to JSON/TSV/Markdown/SQL, column statistics, and generated TypeScript or SQL schemas
Convert cURL to fetch, axios, ky, XHR, Node.js with syntax highlighting
Compare text with syntax highlighting, auto-diff, stats, and patch export
Browse documentation with search, offline support, and syntax highlighting
Hash text with MD5, SHA-1, SHA-256, SHA-512, BLAKE2b, BLAKE3 and bulk generation
Validate HTML with live preview, clickable problems, accessibility rules, heading outline, and formatting
View, convert, resize, and optimize images with format conversion and metadata
Live validation against a JSON Schema with a clickable problems list, 7 templates, inference, sample generation, and strict mode
Validate, format, minify, sort keys, path query, tree and table views for JSON
Decode JWT with header, payload, signature verification, and expiry check
Parse and filter logs with syntax highlighting, timestamps, levels, and export
Edit Markdown with live preview, syntax highlighting, and formatting
Create and edit Mermaid diagrams with live preview and export
Create and manage notes with rich text editing and organization
Manage and use prompt templates with variables and quick access
AST codemods with diff preview (12 transforms, JS/TS)
Test regex with match highlighting, replace, split, lookahead/lookbehind, and flags
Manage code snippets with tagging, search, and quick insertion
Edit text and source files with syntax highlighting and disk reload detection
Timestamps with presets, date picker, timezone, day/week info
Compile TypeScript to JavaScript with live type checking
URL encode/decode with recursive and line-by-line modes, swap, and parsed parts
Generate v1/v4/v5/v7 UUIDs with validation, parsing, and bulk export
Live XML validation with tree, JSON and XPath panes beside the editor; format, minify, copy
Live YAML validation with tree and JSON panes beside the editor; multi-document streams, format, sort keys, compact
Tool descriptions are UI-focused summaries (30-100 chars) rather than LLM-optimized docstrings. None explain WHEN to use the tool, WHAT it returns, or dependencies.
No output schemas documented. LLMs cannot infer what fields each tool returns, breaking downstream tool chaining and forcing agents to guess at response structure.
Tool definitions appear to be React component registrations in src/app/tool-registry.ts, not MCP tool definitions. No explicit MCP server implementation visible despite rmcp dependency in Cargo.toml.
No error handling guidance visible. Tools lack recovery hints, error categorization (retryable vs fatal), or actionable error messages. Agents cannot self-correct on failures.