Build, connect, and execute with LitEngine MCP.
LitEngine implements the standard Model Context Protocol (MCP) specification to provide AI models with deterministic access to UK corporate registries, operational CRMs, and email dispatchers.
JSON-RPC 2.0 Streamable HTTP Transport
LitEngine is built on the official Model Context Protocol transport specification (spec versions 2025-06-18, 2025-03-26, 2024-11-05).
Per the MCP transport spec, requests are answered with Content-Type: application/json. Every request carries its own authentication context, eliminating fragile websocket drops.
Enforces origin validation against hostile browser attacks, allowing authenticated calls from Claude server-to-server or authorized web clients.
{
"protocol": "Model Context Protocol (MCP)",
"jsonrpc": "2.0",
"supportedVersions": [
"2025-06-18",
"2025-03-26",
"2024-11-05"
],
"transports": [
"HTTP POST (stateless)",
"stdio (cli runtime)"
],
"authentication": [
"OAuth 2.0 Bearer Token",
"URL-Secret SHA-256 (60-day grant)"
],
"safetyGates": [
"Human Approval Queue",
"6-digit one-time confirmation codes"
]
}Official connection front doors
LitEngine exposes two production connection routes to accommodate both modern OAuth connectors and token-based environments.
Claude OAuth Front Door
The primary endpoint used in Claude's connector dialog. If called without credentials, it returns an authenticated 401 with standard WWW-Authenticate pointing to the discovery metadata trailhead.
Authorization: Bearer <access_token>Workspace URL Token Endpoint
Designed for clients that accept a single connection URL without custom header configuration. The token is hashed with SHA-256 and expires automatically after 60 days.
Claude Desktop Configuration
Add this to your local claude_desktop_config.json
{
"mcpServers": {
"litengine": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-fetch",
"https://litengine.in/api/mcp/s/YOUR_WORKSPACE_TOKEN"
]
}
}
}Verified MCP Tool Catalog
Every tool below is implemented in production and workspace-isolated. AI clients discover these tools dynamically via tools/list.
Corporate Intelligence & Distress Screening
distress_screen_search()JSON-RPC 2.0Searches UK Companies House filings, parses XBRL accounts, and calculates net asset / working capital distress scores.
{ "sectorCode": "SIC 41202", "minAgeYears": 3, "distressOnly": true }find_contact()JSON-RPC 2.0Retrieves a contact and their associated company profile, statutory status, and outstanding charges.
{ "query": "patric.cassidy@cassidygroupltd.com" }CRM Pipeline & Relationship Queries
who_is_waiting()JSON-RPC 2.0Surfaces high-priority contacts who reached out or require immediate advisor follow-up.
{ "limit": 10 }list_hot_leads()JSON-RPC 2.0Lists active leads scored by high distress indicators, open email signals, and response state.
{ "stage": "qualified", "limit": 25 }pipeline_summary()JSON-RPC 2.0Returns aggregated counts of contacts across all stages (New, Qualified, Sequence Active, Review, Closed).
{}search_contacts()JSON-RPC 2.0Performs workspace-scoped search with tag filtering, domain matching, and status constraints.
{ "filter": { "tag": "priority_distress", "hasEmail": true } }Intelligent Sequences & Approval Gates
create_sequence()JSON-RPC 2.0Creates a draft multi-step email sequence with custom step delays and tone guardrails.
{ "name": "Distress Follow-up", "stepCount": 3, "intervalDays": 4 }preview_sequence()JSON-RPC 2.0Generates rendered previews of all emails for an approved list, including verified financial figures.
{ "sequenceId": "seq_9481a" }approve_sequence()JSON-RPC 2.0Requires a 6-digit server-minted verification code to release messages into the approved send queue.
{ "sequenceId": "seq_9481a", "confirmCode": "582914" }review_outbox()JSON-RPC 2.0Inspects currently pending messages in the operator approval queue.
{ "status": "pending_approval" }Request Sandbox API Tokens
Connect your development team to test corporate XBRL parsing and CRM execution directly in your local environment.