Lead Scorer

Personal CRM memory from Cursor

Lead Scorer exposes its whole outbound motion as an MCP server. Point Cursor at it and personal, relationship and company memory becomes something you ask for in a sentence instead of something you click through. Useful when your ICP is easiest to describe next to the code: the agent reads your product and briefs itself.

Bring what your agent already knows about you and your relationships into a private, versioned Markdown memory that every Lead Scorer action can reuse.

Connect Cursor

Bearer MCP API key generated in Lead Scorer settings, the manual path for clients without a working OAuth flow. Runs in the Cursor editor, from a JSON config file.

  1. 1

    Create an MCP API key in Settings and copy it before closing the dialog.

  2. 2

    Replace the placeholder in Cursor's MCP configuration.

Server URL

https://mcp.lead-scorer.com/mcp
189 tools · included on the free plan
~/.cursor/mcp.json
{
  "mcpServers": {
    "lead-scorer": {
      "url": "https://mcp.lead-scorer.com/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_MCP_API_KEY>"
      }
    }
  }
}
Cursor MCP documentation

Three prompts to start with

Paste one into Cursor once the server is connected. No tool names required: it picks them.

  • Read what Lead Scorer and this chat remember about me, identify stale or contradictory sources, and propose the corrected canonical memory for review.
  • Search my companies for Jules, extract the durable facts from this selected Obsidian mission note, save them into its company memory with submit_memory_facts, then show me the resulting canonical revision and Context Pack.
  • List my memory sources waiting for analysis with list_pending_memory_sources, extract the durable facts from each one, and save them with submit_memory_facts.

What Cursor can do on personal, relationship and company memory

  • Read and update your canonical my-memory.md without exposing another user's document.
  • Compare that memory with dated, source-attributed workspace evidence before correcting identity, role or positioning.
  • Keep one separate Markdown memory per relationship, visible from the matching lead record.
  • Keep company history in its own Markdown memory, visible from the matching company record and reusable for every related lead.
  • Append a selected fact from ChatGPT, Claude or the CRM with its raw source and browse the immutable revision history.
  • Compile a bounded Context Pack for a lead or operation from personal memory, relationship memory and verified CRM overlays.
  • Let your own AI read the context you saved from ChatGPT, Claude, transcripts or notes and write every grounded fact back in a single revision, at no AI cost.
  • Audit every imported fact through its source and immutable history; direct user-authored edits stay free and versioned.

Tools involved

  • get_my_memory
  • get_my_workspace_context
  • update_my_memory
  • list_relationship_memories
  • get_relationship_memory
  • list_memory_revisions
  • update_relationship_memory
  • remember_relationship_fact
  • archive_relationship_memory
  • list_company_memories
  • get_company_memory
  • update_company_memory
  • remember_company_fact
  • archive_company_memory
  • list_memory_sources
  • get_memory_source
  • archive_memory_source
  • list_pending_memory_sources
  • submit_memory_facts
  • compile_context_pack
  • list_memory_proposals
  • review_memory_proposal
  • review_memory_proposals

Where you stay in the loop

Your agent may write only your own memory. Your AI does the analysis, so memory imports cost no AI budget; every fact keeps a verbatim excerpt of its source, and memory never approves or sends outreach.

What is Cursor?

Cursor is an AI code editor that speaks MCP. Add one block to its config and the same agent that writes your product can also run the outreach for it.

Built by Anysphere.

Start free

The CRM and MCP access are free forever, no credit card. Connect Cursor, ask for personal, relationship and company memory, and approve what comes back.