MCP.so
Sign In

Proxenio MCP Server

@proxenio

About Proxenio MCP Server

MCP server for Proxenio — verified intent network for professional deal-making. Agent discovery, match retrieval, and deal acceptance via Model Context Protocol.

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "proxenio": {
      "command": "npx",
      "args": [
        "-y",
        "@proxenio-tech/mcp-server"
      ]
    }
  }
}

Tools

4

Get information about the Proxenio platform and its agent API. Proxenio is a verified intent network for professional deal-making. AI agents can access their human principal's matches through API keys with inherited trust tiers. This tool requires no authentication. Use it to understand the platform before connecting. Returns: Platform description, trust model, available capabilities, authentication method, and rate limits. Examples: - Use when: "What is Proxenio?" or "How do I connect to Proxenio?" - Use when: Agent needs to understand what Proxenio offers before authenticating - Don't use when: You already know about Proxenio and just need to read matches

Configure the API key for authenticating with Proxenio's agent API. This must be called before using proxenio_get_matches or proxenio_accept_match. The API key is generated by a verified human principal at https://www.proxenio.ai/agents. Keys are 36 characters, start with 'prx_', and inherit the principal's trust tier. Args: - api_key (string): The full API key starting with 'prx_' (36 characters) Returns: Confirmation that the key was set, or an error if the format is invalid. Examples: - Use when: Setting up Proxenio access for the first time - Use when: Switching to a different principal's API key - Don't use when: Key is already configured and working

Read the principal's verified professional matches from Proxenio. Returns matches sorted by match score (descending). Each match includes counterparty profile, trust tier, PRO score, intent, verification status, and confirmed outcome count. Requires: API key must be set via proxenio_set_api_key first. The agent sees exactly what the human principal sees — same verification gates, same tier gates, same minimum score (40). Matches below 40 are never returned. Args: - filter_type ('all' | 'top' | 'high' | 'standard'): Filter by match quality tier (default: 'all') - filter_status ('all' | 'pending' | 'accepted'): Filter by match status (default: 'all') - min_score (number, 40-100): Minimum match score threshold (default: 40) Returns: Principal info, list of matches with full counterparty profiles, trust signals, and rate limit status. Examples: - Use when: "Show me my matches" or "What professional connections are available?" - Use when: "Find my best matches" -> use filter_type='top' - Use when: "Which introductions are pending?" -> use filter_status='pending' - Don't use when: API key has not been set yet (use proxenio_set_api_key first) Error Handling: - Returns auth error if API key is invalid or revoked - Returns rate limit error if 60 requests/hour exceeded - Returns empty list if principal has no matches (profile may be incomplete)

Accept an introduction request on behalf of the human principal. This creates a deal between the principal and the counterparty. The principal must be the receiving party (user_b) — agents cannot accept introductions that the principal initiated. IMPORTANT: This is an action with real consequences. It connects two professionals and creates a deal container. Only accept matches that align with the principal's stated intent. Requires: API key must be set via proxenio_set_api_key first. Args: - match_id (string, UUID): The match ID to accept. Get this from proxenio_get_matches results. Returns: Confirmation with counterparty details, the new deal ID, and rate limit status. Guard rails enforced by the API: - Principal must be a participant in the match - Principal must be the receiving party (user_b), not the requester - Match must have an active status - An introduction request must exist and be in pending/awaiting_consent state - Agent cannot accept the principal's own outgoing requests Examples: - Use when: Principal wants to accept a specific pending introduction - Use when: Agent decides a match aligns with principal's intent and should be accepted - Don't use when: You haven't checked the match details first (use proxenio_get_matches) - Don't use when: The match was initiated by the principal (agent can only accept incoming) Error Handling: - 400: Match not in correct state, no intro request exists - 403: Principal is not authorized (e.g., they are the requester, not receiver) - 404: Match ID not found - 429: Rate limit exceeded

Overview

What is Proxenio MCP Server?

An MCP server that connects AI agents in Claude Desktop, Cursor, VS Code, and other MCP-compatible clients to the Proxenio verified intent network. It enables agents to discover platform information, set API keys, retrieve professional matches, and accept introduction requests using the same matching engine and verification rules as the human principal.

How to use Proxenio MCP Server?

Install globally with npx @proxenio-tech/mcp-server. Configure it as an MCP tool in your client’s settings (e.g., claude_desktop_config.json). The server exposes four tools: proxenio_discover, proxenio_set_api_key, proxenio_get_matches, and proxenio_accept_match. For remote deployment, set TRANSPORT=http PORT=3001.

Key features of Proxenio MCP Server

  • Four MCP tools for Proxenio’s matching engine
  • Agents inherit the human principal’s trust tier
  • Match filtering by quality, status, and minimum score
  • Accept introductions with guard rails (must be receiving party)
  • Rate limiting: 60 requests/hour per key, 3 keys per user
  • HTTP transport option for multi-client or cloud deployment

Use cases of Proxenio MCP Server

  • An AI assistant in Claude Desktop discovers Proxenio platform details without authentication
  • A developer sets an API key via proxenio_set_api_key to enable authenticated actions
  • An agent reads a principal’s verified professional matches using proxenio_get_matches
  • An agent accepts a pending introduction on behalf of the principal with proxenio_accept_match
  • A cloud-hosted agent exposes the same tools over HTTP for multiple clients

FAQ from Proxenio MCP Server

What tools does Proxenio MCP Server provide?

Four tools: proxenio_discover (no auth required), proxenio_set_api_key, proxenio_get_matches, and proxenio_accept_match. See the Tools Reference section for detailed parameters.

What are the prerequisites for using Proxenio MCP Server?

Node.js 18+, a Proxenio account with a verified email and completed profile, and an API key generated at proxenio.ai/agents.

How do I install Proxenio MCP Server?

Run npx @proxenio-tech/mcp-server. Alternatively, clone the repository and build from source.

What are the rate limits?

60 requests per hour per API key, with a maximum of 3 keys per user (giving 180 requests/hour total). Rate limit headers are included in all responses.

How does the trust model work?

Agents inherit their human principal’s trust tier at request time. Trust tiers range from 0 (Unverified) to 4 (Proven). All engine verification gates still apply, and counterparties see a transparency badge indicating an AI agent is acting on behalf of the named principal. Only humans can log outcomes and confirm deals.

Frequently asked questions

What tools does Proxenio MCP Server provide?

Four tools: `proxenio_discover` (no auth required), `proxenio_set_api_key`, `proxenio_get_matches`, and `proxenio_accept_match`. See the Tools Reference section for detailed parameters.

What are the prerequisites for using Proxenio MCP Server?

Node.js 18+, a Proxenio account with a verified email and completed profile, and an API key generated at [proxenio.ai/agents](https://www.proxenio.ai/agents).

How do I install Proxenio MCP Server?

Run `npx @proxenio-tech/mcp-server`. Alternatively, clone the repository and build from source.

What are the rate limits?

60 requests per hour per API key, with a maximum of 3 keys per user (giving 180 requests/hour total). Rate limit headers are included in all responses.

How does the trust model work?

Agents inherit their human principal’s trust tier at request time. Trust tiers range from 0 (Unverified) to 4 (Proven). All engine verification gates still apply, and counterparties see a transparency badge indicating an AI agent is acting on behalf of the named principal. Only humans can log outcomes and confirm deals.

Comments

More Other MCP servers