Official Python SDK and CLI for the DiscoLike API - a search engine for the business web providing company discovery, enrichment, contact finding, and matching capabilities
This SDK/CLI presents critical definition quality gaps that would require significant remediation before production use. While tool names follow verb-noun patterns, schemas are partially present but inconsistent, and descriptions lack depth. Most critically, the implementation is STDIO-only, which hard-caps protocol readiness at 50 and makes this unsuitable for hosted/remote agent clients. Of 9 tools, only 3-4 have complete input schemas visible in the source. Parameter descriptions are present but generic. No output schemas are documented. Error handling and recovery guidance are absent. The server appears to be a Python SDK wrapper around an external API (DiscoLike), not a native MCP implementation.
Retrieve month-to-date usage metrics including request count, records processed, and spend
Retrieve comprehensive company profile data including firmographics, industry classification, business model, and description for a single domain
Match a bulk set of contacts to DiscoLike persona IDs with match scores using name, email, domain, and optional company name and country
Search for contacts at companies matching specified seniority, department, title, and other criteria
Count the total number of companies matching search criteria without retrieving the full result set
Submit an AI research job to run a custom query over discovered domains and return structured answers per company with optional web search
STDIO-only transport: not remotely accessible, incompatible with hosted MCP clients. This is a hard architectural limitation.
No output schemas documented. Tools return data but LLMs cannot know the structure of responses, forcing them to guess at response field names and types for chaining downstream calls.
No error handling guidance. Error responses do not tell the LLM what went wrong, how to recover, or whether to retry. A bare exception or API error code gives the agent nothing actionable.
Tool descriptions lack depth and dependency hints. Many descriptions (e.g., 'Retrieve comprehensive company profile data...') lack context on when to call the tool instead of related ones, and do not explain prerequisites or typical workflows.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
Discover lookalike companies matching search criteria (domain similarity, ICP prompts, phrase matching, filtering by country and employee range)
Batch find work email addresses for people given first name, last name, and company domain with status and validation results
Open a new DiscoLike account for a person without requiring an API key (public endpoint)
Dot-separated tool names (company.data, contacts.search, discogen.process) are unconventional for MCP. Standard pattern is underscore-separated (company_data, contacts_search) for better parsing and LLM readability.
Parameter descriptions are inconsistent. Some (e.g., 'ISO-2 country codes or region aliases') are specific; others ('Company domains to search contacts at') are vague about format, count, or limitations.
No pagination guidance documented. Tools like 'discover' and 'contacts.search' accept 'max_records' but do not document how to handle large result sets, whether pagination is supported, or how to iterate through results.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Only the 'signup' tool is identified as WRITE in metadata, but the tool definitions themselves lack structured hints for the LLM.
account.usage accepts no parameters but description does not explain what period 'month-to-date' covers or what metrics are returned.