Loading ToolBench
Whissle MCP Server - provides access to Whissle API endpoints for speech-to-text, translation, and summarization services
Two tools with complete input schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. Both tools have cost warnings and error handling, but descriptions lack LLM-optimized structure. Parameters like 'boosted_lm_words' and 'max_speakers' have minimal descriptions. Output schema is not formally documented, only mentioned in docstring as 'TextContent'. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Error handling is present in code but not reflected in tool definitions.
Convert speech to text with a given model and save the output text file to a given directory. Directory is optional, if not provided, the output file will be saved to $HOME/Desktop. ⚠️ COST WARNING: This tool makes an API call to Whissle which may incur costs. Only use when explicitly requested by the user. Args: audio_file_path (str): Path to the audio file to transcribe model_name (str, optional): The name of the ASR model to use. Defaults to "en-NER" timestamps (bool, optional): Whether to include word timestamps boosted_lm_words (List[str], optional): Words to boost in recognition boosted_lm_score (int, optional): Score for boosted words (0-100) output_directory (str, optional): Directory where files should be saved. Defaults to $HOME/Desktop if not provided. Returns: TextContent with the transcription and path to the output file.
Convert speech to text with speaker diarization and save the output text file to a given directory. Directory is optional, if not provided, the output file will be saved to $HOME/Desktop. ⚠️ COST WARNING: This tool makes an API call to Whissle which may incur costs. Only use when explicitly requested by the user. Args: audio_file_path (str): Path to the audio file to transcribe model_name (str, optional): The name of the ASR model to use. Defaults to "en-NER" max_speakers (int, optional): Maximum number of speakers to detect
Output schema not formally documented. Tool descriptions mention 'TextContent with the transcription and path to the output file' but no structured schema is provided. LLMs cannot plan downstream operations without knowing exact output fields.
Parameter descriptions are minimal or missing context. 'boosted_lm_words' (array of strings) lacks guidance on expected format, count limits, or use cases. 'max_speakers' has no range or default behavior documented. Descriptions should be 50-200 chars with actionable constraints.
No tool annotations (destructiveHint, idempotentHint, readOnlyHint). Both tools write files to disk and call external APIs with cost implications, but lack annotations to signal this to agents. Agents cannot reason about side effects or retry safety.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | <=2025-11-25 | v2 |
Error handling is implemented in code (handle_api_error function with retry logic, HTTP status codes, file validation) but not exposed in tool definitions. LLMs see only the happy path and cannot anticipate or handle failures gracefully.
Parameter 'output_directory' defaults to null with fallback to $HOME/Desktop in description, but schema shows default: null. Inconsistency between schema default and documented behavior. Should either set schema default to '$HOME/Desktop' or clarify the fallback in description.