Crawdar MCP - Qualified Prospect Research for Agents¶
Remote MCP server (Streamable HTTP) - Crawdar converts "find companies that match this" into qualified prospects: each result carries why the company matches, criterion-level decisions, a classified public contact route and the supporting sources. A versioned JSON API and a browser-capable interface sit alongside the MCP surface.
Server type: Remote (Streamable HTTP)
Auth: Tool surface served anonymously (probe-verified); search capacity via the vendor's plans
Endpoint: https://crawdar.com/api/mcp (probe-verified live; 9 tools)
Tools: 9 (research_businesses, search_businesses, sandbox_businesses, lead-search job lifecycle, explain_crawdar)
Pricing: Plans at crawdar.com
Category: Lead Generation & Web Scraping
Built by: Crawdar (crawdar.com)
Why This Matters for Operators¶
Prospecting tooling falls into two bad shapes: raw page scrapes the agent must re-interpret, and closed CRMs that hide why a lead was surfaced. Crawdar sits in the middle: every result names the match reasons and cites the sources, so the operator can audit the recommendation instead of trusting a black box - and the sandbox tool lets agent loops be tested on deterministic fictional businesses before touching live capacity.
The durable job model is the workflow win: start_lead_search returns a private job with progress, export, retry and cancel URLs, so long searches survive the conversation instead of blocking the agent call.
Tools & Capabilities¶
| Tool | What it does |
|---|---|
| research_businesses | One-call natural-language search with compact or full output |
| search_businesses | Structured search with explicit target, geography, qualifier and field selection |
| sandbox_businesses | Deterministic fictional results for contract and agent-loop testing |
| start_lead_search | Creates a private async search job with progress, status, export, retry and cancel URLs |
| get_lead_search | Reads job progress or cursor-paginated result pages |
| refine_lead_search | Creates a new job by changing the brief or criteria, preserving the original |
| retry_lead_search | Replaces a failed or window-expired job; the original stays available for audit |
| cancel_lead_search | Cancels a queued or running job; completed and failed jobs remain immutable |
| explain_crawdar | Returns result semantics, limits, interface URLs and safe operating guidance |
Installation¶
Point any Streamable HTTP client at the endpoint. Start with sandbox_businesses to validate schemas and agent loops without consuming live search capacity.
{
"mcpServers": {
"crawdar": {
"type": "http",
"url": "https://crawdar.com/api/mcp"
}
}
}
Configuration¶
The probe served the full 9-tool surface anonymously; live search capacity is governed by the operator's Crawdar plan. The explain_crawdar tool documents result semantics and limits from inside the client.
Business Relevance¶
- Sales teams replace manual list-building with sourced, reason-carrying prospect searches.
- RevOps teams run durable async lead searches that survive the conversation.
- Agencies test prospect pipelines in the sandbox before spending live capacity.
- Founders get qualified prospects with contact routes for first outbound motion.
Integration with CorpusIQ¶
Crawdar finds the prospects; CorpusIQ verifies the business. A composed workflow: the agent pulls qualified prospects from Crawdar, enriches the shortlist through CorpusIQ's business data connectors, and tracks pipeline-to-revenue through the CRM and Stripe data - sourcing, qualification and attribution in one loop.
Limitations¶
- Brand new listing (mcpservers.org /all, Sep 2026); no public repository located, stars n/a.
- Result quality and capacity depend on the Crawdar plan tier.
- Contact routes are public-facing classifications, not personal-contact data.
- The sandbox is deterministic but fictional; it validates loops, not real data quality.