Loading ToolBench
MCP server for audio transcription using Gemini models via OpenRouter
Strong tool naming (verb_noun pattern) and comprehensive parameter descriptions. All 6 tools have clear, detailed descriptions (150-250 chars each). Input schemas are well-structured with type definitions and descriptions for all parameters. However, output schemas are not documented, LLMs cannot predict return structure. Error handling guidance is absent. No tool annotations (readOnlyHint, destructiveHint). Parameter descriptions are excellent but lack explicit format constraints (e.g., no enum for model choices despite mentioning 'lite' vs 'flash'). Composition is sound: tools are single-purpose and chainable via file_url/file_content patterns.
Lists available transcription presets from the Text-Transformation-Prompt-Library. Each preset is categorized as either a "style" (modifies tone/voice without changing structure) or a "format" (restructures content into a specific document type). Use these with the transcribe_with_preset tool.
Transcribes an audio file using Gemini via OpenRouter. Returns a lightly edited transcript with filler words removed, verbal corrections applied, punctuation added, and paragraph breaks inserted. Large files are automatically compressed. Supports MP3, WAV, OGG, FLAC, AAC, AIFF, and many more formats (auto-converted). This is the recommended default tool.
Transcribes an audio file using Gemini via OpenRouter with a user-defined custom prompt. Provides full control over how the model processes and formats the transcription. Use this when you need specific transcription instructions not covered by other tools.
Transcribes an audio file and formats the output as a specific document type. Accepts any freeform format description. Use this when you want a quick ad-hoc format without browsing presets. For curated, high-quality formatting, use transcribe_with_preset instead.
Output schemas not documented. LLMs cannot predict return structure (transcript format, fields, pagination). Agents must guess whether results are paginated or what fields to extract.
Model parameter accepts free-form strings ('lite', 'flash', or full OpenRouter model IDs) but lacks enum constraint. LLMs may hallucinate invalid model names. Should define enum: ['lite', 'flash'] or document regex pattern.
No error handling guidance. Tool descriptions do not explain what to do if transcription fails, file is inaccessible, or API quota is exceeded. Agents have no recovery path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
Transcribes an audio file using Gemini via OpenRouter. Returns a verbatim transcript with NO cleanup - preserves filler words, false starts, and repetitions exactly as spoken. Use this when you need exact speech-to-text without any editing.
Transcribes audio and transforms the output using a curated preset. Presets are divided into two categories: **Styles** (modify tone/voice): formal, informal, academic, business, journalistic, assertive, flamboyant, minimalist, dejargonizer, simplify, victorian, shakespearean, etc. **Formats** (restructure into document type): blog_outline, business_email, meeting_minutes, note_to_self, to_do_list, tech_documentation, feature_request, bug_report, cover_letter, resume, newsletter, development_prompt, etc. Use list_transcription_presets to browse all 200+ available presets with category filters.
No tool annotations. Tools lack readOnlyHint (all are read-only), destructiveHint (none are destructive), or idempotentHint. Agents cannot reason about retry safety or side effects.
Mutually exclusive parameters (file_content vs file_url vs ssh_host+ssh_path) not explicitly documented as such. LLMs may pass multiple conflicting parameters, causing ambiguous failures.