Loading ToolBench
A voice-to-text formatting service that converts raw transcripts into styled formats (bullet points or prose)
This MCP server has a single tool (format_dictation) with severe quality gaps. The tool lacks a proper MCP server registration structure, it's a bare Python class with no evidence of MCP protocol integration. The input schema is visible but incomplete; the output schema is undocumented. The tool description is present but vague and lacks actionable guidance. Parameter descriptions exist but are minimal. No error handling, no input validation, no security considerations, and no evidence of MCP transport or protocol compliance.
Formats raw voice transcripts into styled text with configurable output format
No MCP server registration or protocol implementation visible, source shows only a Python class, not an MCP-compliant server with transport, capabilities, or tool registration.
Output schema completely undocumented, tool returns {'formatted_text': str, 'word_count': int} but no schema is declared. LLMs cannot plan downstream operations or validate responses.
Tool description is generic and lacks context, 'Formats raw voice transcripts into styled text with configurable output format' does not explain WHEN to use it, what styles are valid, or what happens when an invalid style is passed.
Parameter 'target_style' lacks enum constraint, description says 'e.g., bullet_points or prose' but code only handles bullet_points and defaults to prose. LLMs will pass undocumented values like 'markdown', 'summary', etc.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
No error handling or validation, tool will silently accept any string for target_style and default to prose if unrecognized. No guidance on recovery if input is invalid.
Tool logic is dummy, formatter only takes first 3 words and appends ' details...' placeholder. This is not production-grade; real formatting logic is missing.
Parameter descriptions are too brief, 'The raw transcribed text from voice input' and 'The desired output format (e.g., bullet_points or prose)' are under 72 characters average; the second includes example values that LLMs will treat as the only valid options.