MCPWorks with Claude Code, Cursor & Codex
GTM infrastructure for your AI agent
Connect your coding agent over MCP and it finds leads, writes sequences, and runs LinkedIn outbound end to end from the client you already work in.
you find 25 seed-stage RevOps leads in NYC and start a sequence
→ rilo.search_people() 25 matches · enriched
→ rilo.get_person_profile() 25 enriched · verified
→ rilo.create_sequence() 3 steps · personalized
→ rilo.send_connection_request() queued · 25
Sequence live. I'll surface replies in your inbox and keep tuning.
Quick connect
Connect in under a minute
One command configures Claude Code, Cursor, or Codex. The client handles sign-in afterward when its remote MCP support allows it.
$npx rilogtm connect
Writes the Rilo URL. Supported clients prompt for browser sign-in; API tokens remain the headless fallback.
Clients
Pick your client
Each guide walks the exact setup. Same endpoint, same tools, whichever you use.
CLI
Claude Code
One command writes the URL; Claude Code handles sign-in.
MCP
Cursor
One command writes the MCP config.
CLI
Codex
One command writes the MCP config.
claude.ai
claude.ai / Claude Desktop
Open Claude with Rilo prefilled, then sign in with Google.
Connectors
ChatGPT
Connector OAuth is not production-verified yet.
CLI
OpenClaw
One command writes the MCP config.
Endpoint
HTTP API
Same operations over plain HTTP, for your own code.
Shell
CLI
Wire up any client from your shell.
Tool reference
What your agent can do
Around 50 tools across seven areas, every one documented with parameters and real sample requests. Your agent finds leads, sends outreach, books meetings, and keeps improving as it learns.
15 tools
Search & enrichment
14 tools
Sequences
9 tools
LinkedIn actions
6 tools
Inbox
5 tools
Learning
4 tools
Connect
5 tools
Observability