Loading ToolBench
MCP server for interacting with the Trykitt API to find and verify email addresses
Three tools with complete schemas and descriptions, but significant gaps in parameter clarity and error handling. Tool names follow verb_noun convention (get_, find_, verify_). Descriptions are detailed (150-400 chars) but lack actionable error guidance. Parameters are well-typed with Zod schemas and mostly described, but optional parameters lack clarity on when to use them. No output schema documentation. Error handling returns generic failures without recovery guidance. Missing idempotency hints and permission declarations.
Initiates a real-time job (realtime: true is forced) to find a person's email address via the POST /job/find_email endpoint (https://api.trykitt.ai/job/find_email). Required parameters: 'fullName' (provide the name as-is from LinkedIn/directories, do not clean) and 'domainOrWebsite' (provide the company's email domain like 'trykitt.ai' or website URL like 'https://trykitt.ai'; domains are preferred over websites for better results). Optional parameter: 'linkedinStandardProfileURL' (recommended for accuracy, especially if the full name is obscured on LinkedIn). Results are returned directly in the response; callback URLs are not used by this implementation. Requires x-api-key header.
Retrieves the status or result of a previously submitted Trykitt job using the GET /job endpoint. Requires the 'job_id' parameter. Sends requests to https://api.trykitt.ai/job?id={job_id} with the configured x-api-key header.
Initiates a real-time job (realtime: true is forced) to verify if an email address is valid via the POST /job/verify_email endpoint (https://api.trykitt.ai/job/verify_email). Required parameter: 'email' (the address to verify). Optional parameter: 'treatAliasesAsValid' (boolean, defaults to true in this implementation) which determines if alias addresses should be considered valid. Results are returned directly in the response; callback URLs are not used by this implementation. Requires x-api-key header.
No output schema documentation. Tools return JSON.stringify(result) without documenting what fields the LLM should expect. Agents cannot plan downstream calls or extract structured data reliably.
Error responses lack recovery guidance. Failures return generic text like 'Failed to retrieve job status' without suggesting next steps (e.g., 'Check job_id format' or 'Retry in 5 seconds'). LLMs cannot self-correct.
Optional parameters (callbackType, callbackURL, companyName) lack guidance on when to use them. Description states they are 'Optional' but does not explain when they improve results or are required for specific use cases.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
No idempotency or permission declarations. Tools do not state whether repeated calls with same input produce same result, or what permissions are required. Agents cannot safely retry or understand scope.
API key exposed via environment variable without validation. TRY_KITT_API_KEY defaults to 'YOUR_TRYKITT_API_KEY' string if env var missing, causing silent failures instead of clear error on startup.