Docs
MCP / API

MCP / API

Connect Claude, Cursor, n8n, or your own agent to the KithFlow MCP server.

MCP / API

KithFlow exposes a Model Context Protocol (MCP) server with 11 tools. Any MCP-compatible client — Claude.ai, Cursor, n8n, or your own Python/Node agent — can connect and operate your entire sales pipeline programmatically.

What is MCP?

MCP (Model Context Protocol) is an open standard for exposing tools to AI models over a consistent interface. Instead of building custom integrations per model or client, KithFlow exposes one MCP server and any compatible client can connect to it with an API key.

Server Details

PropertyValue
Server URLhttps://api.kithflow.com/mcp
TransportServer-Sent Events (SSE)
AuthAPI key (passed as a tool parameter)
ProtocolMCP spec v1.0+

Authentication

Generate an API key in Settings → API Keys. Pass it as the api_key parameter in every tool call alongside your org_id.

API keys are SHA-256 hashed at rest and scoped to your organisation — no cross-org data is accessible.

All 11 Tools

Prospect Discovery

ToolDescription
search_prospectsFind companies by keyword + location using Google Maps and/or PDL
enrich_prospectScrape a company's website, find a contact email via Hunter.io, generate an icebreaker

Sequence Management

ToolDescription
enroll_in_sequenceEnroll a prospect in a named email sequence
list_sequencesList all email sequences available in the org
list_active_enrollmentsGet all active sequence enrollments and their current step
pause_enrollmentPause outreach for a specific prospect

UK Companies

ToolDescription
search_uk_companiesSearch the Companies House database by name, SIC code, or postcode

Battlecard

ToolDescription
get_prospect_dossierFetch the battlecard (dossier) for a given prospect ID
generate_dossierTrigger battlecard generation for a prospect (returns immediately; poll get_prospect_dossier for status)

Intent Engine

ToolDescription
create_targeting_ruleCreate a new targeting rule with keywords, location, cadence, and sequence
list_targeting_rulesList all targeting rules for the org
trigger_targeting_ruleTrigger an immediate one-off run of a specific rule

Connecting Claude.ai

  1. In Claude.ai, go to Settings → Integrations → MCP Servers
  2. Add a new server:
    • Name: KithFlow
    • URL: https://api.kithflow.com/mcp
    • API key: your KithFlow API key
  3. Save — the 11 tools appear in Claude's tool list

You can now give Claude instructions like:

"Find 20 software development companies in Manchester, enrich them, and enroll them in my Q3 sequence."

Claude will chain search_prospectsenrich_prospectenroll_in_sequence automatically.

Connecting Cursor

In your .cursor/mcp.json (or the Cursor MCP settings UI):

{
  "mcpServers": {
    "kithflow": {
      "url": "https://api.kithflow.com/mcp",
      "transport": "sse"
    }
  }
}

Pass your API key and org ID when invoking tools from the Cursor chat.

Connecting n8n

Use the HTTP Request node or an MCP Client node (if your n8n version supports it) to call the KithFlow MCP server.

Example flow:

  1. Trigger — Schedule node (e.g. every Monday 08:00)
  2. KithFlow: trigger_targeting_rule — pass rule_id
  3. Wait — 5 minutes
  4. KithFlow: list_active_enrollments — get new enrollments
  5. Slack: send message — post a summary to your channel

Example: Python Agent

import anthropic
 
client = anthropic.Anthropic()
 
# Claude will use KithFlow MCP tools automatically
response = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    tools=[...],  # loaded from MCP server manifest
    messages=[{
        "role": "user",
        "content": (
            "Trigger my 'London IT Support' targeting rule, "
            "wait for it to complete, then tell me how many "
            "new prospects were enrolled."
        )
    }]
)

Rate Limits

LimitValue
Requests per minute60
search_prospects per hour20
enrich_prospect per hour50

Exceeding a limit returns a 429 Too Many Requests response. Retry after the Retry-After header value.