RevorRevor

REVOR AI MCP

REVOR AI MCP connects company and person discovery, company research, contact discovery, customs trade intelligence, and authorized outreach to AI clients that support MCP. Once connected, you can request work in natural language and let the client choose the appropriate REVOR AI tools.

Server URL

https://revor.ai/api/mcp

This is a remote service; no local server is required. Install it with an API key: create a key on the API Keys page, then store it in the client's private x-api-key header.

To let an AI assistant install it for you, send the assistant the AI installation guide. It contains the API key entry point, header configuration, and post-install verification steps.

Get started

  1. Create or select a REVOR AI API key on the API Keys page.
  2. Add a new MCP service in a client that supports remote MCP.
  3. Enter REVOR AI as the name and https://revor.ai/api/mcp as the server URL.
  4. Add x-api-key: sk-revor-... in the client's private Header or Secret settings.
  5. Save the connection, reload tools, and call revor_whoami to verify the account and permissions.

The client must provide a secure Header or Secret setting. If it does not, use an MCP client that supports custom headers.

Clients may label the entry point as MCP, Tools, Connectors, or Custom Services. The server URL remains the same.

See Connect an MCP client for detailed steps.

You can also send this exact request to an AI assistant that can manage MCP settings:

Text
Read https://revor.ai/docs/mcp/install and install REVOR AI MCP exactly as instructed. Use https://revor.ai/api/mcp and create an API key at https://revor.ai/my-api-keys if needed. Never ask me to send an API key in chat; prefer the client's private x-api-key Secret input. If only a private local configuration on a single-user device is available, write a placeholder and tell me where to replace it locally, ensuring the file is outside source control, shared folders, and cloud sync. After installation, call revor_whoami to verify the connection.

What you can do

  • Create a company or person prospect list from a natural-language target profile and page through matching results.
  • View credits and send-capable connected accounts, then perform a confirmed outreach or LinkedIn engagement action.
  • Research companies, markets, and recent developments from public information.
  • Find business contacts by verified company domain and role.
  • Discover importers or exporters by product, country, and date range, or resolve a known company name before querying counterparties, categories, trends, and country distribution.
  • Create a complete customs trade report.
  • Check asynchronous job progress, results, and credit usage.

See MCP tool reference for the complete list.

Account access and data isolation

MCP uses the REVOR AI account that owns the API key. A client can only use that account's capabilities and retrieve jobs owned by that account. It cannot read another user's data.

You can revoke or rotate an API key at any time. After revocation, the client must be configured with a new valid key before it can continue.

Credits and plans

MCP uses the same capabilities, permissions, and credit rules as the REVOR AI API:

  • Billable tools consume credits according to the corresponding API pricing rules.
  • Customs company-name discovery is currently free.
  • Customs queries settle against rows actually returned; a zero-row result ultimately consumes 0 credits. Job creation can still check and temporarily reserve balance from the requested page_size, then release unused credits at settlement.
  • When a completed job includes billing details, those details are the source of truth.
  • Plan permissions, account balance, and usage limits still apply.

How jobs return results

Creating a prospect list and running some research and customs queries take time. Action tools first return a job or resource ID.

  • succeeded: the result is ready.
  • queued or running: call revor_get_job with the job ID.
  • failed: review the public error code, adjust the input, or retry later.

Reading an existing job does not create another job or charge again for the original operation.

Creating a prospect list returns its ID (webset.id) and a preparation job ID (preparation_job.id). A successful preparation job means that discovery has started, not that all results are complete. Use revor_get_webset for overall progress and revor_list_webset_items to page through results.

Best practices

  • Provide an accurate company name, website domain, date range, and company role.
  • When no company is known, discover companies by HS code or product keywords and target country. Resolve the queryable name before analyzing a specific known company.
  • A single customs query can cover at most one year. Split longer comparisons into separate ranges of no more than one year each.
  • Data-job creation has a separate, relatively low RPM allowance. Create research, contact, and customs jobs serially instead of submitting several creation calls in parallel.
  • Ask the client to continue checking revor_get_job until the job reaches a final state.
  • Never ask for or accept an API key in chat. Prefer the client's Secret store. A plaintext key is acceptable only in a private local configuration on a single-user device with restricted file access; never put one in a shared, cloud-hosted, or synchronized configuration.