MCP server
Lead Scorer's entire outbound motion — finding real companies, scoring them, enriching them, writing the messages — is exposed as 131 tools your assistant can call. Connect it once and outreach becomes something you ask for in a sentence, in the assistant you already have open.
Model Context Protocol is an open standard for letting an AI assistant call someone else's software. A server publishes a list of tools; your assistant reads that list and decides which ones to call while it talks to you. Adding the Lead Scorer server to Claude or ChatGPT is roughly like giving it a login to your CRM — except you decide, per action, what it is allowed to finish.
Pick your client, follow two or three steps, and sign in with your Lead Scorer account. The tools appear in your next message.
Server URL
https://mcp.lead-scorer.com/mcpOAuth — sign in with your Lead Scorer account, no key to copy.
Name: Lead Scorer
Remote MCP URL: https://mcp.lead-scorer.com/mcpNo account yet? Continue anyway: the OAuth browser flow includes account creation and brings you back to your agent after consent. The CRM and MCP access are free, no credit card.
Future tool updates use this same endpoint. There is nothing to reinstall; open a new conversation or refresh the connector only if your current conversation still shows an older cached tool list.
Agent-native setup
The plugin adds the MCP and Skills. OAuth opens Lead Scorer; sign in or create the account, then approve access.
Inside the agent, the commercial catalog is deliberately limited to Pro monthly and one-time credit top-ups.
Connect email or LinkedIn in a hosted flow, create the ICP and campaign, then review every draft before launch.
No tool names to memorise. These are the kinds of sentences that already work.
Nine categories, 131 tools. Everything below ships today — the tool names are the real ones your agent will call.
Create the account during OAuth, read a single readiness snapshot, subscribe to Pro monthly, buy one-time credit top-ups, poll the hosted payment and start email or LinkedIn connection flows without leaving the agent conversation.
Use your assistant’s native web search to discover companies and freelance professionals, then save dated source links and normalized recruiter data in Lead Scorer. Search your saved talent pool by role, skills, availability, rate, location and source without guessing missing facts.
Create a company without inventing a LinkedIn handle, add its verified LinkedIn page later, then enrich it with business context. Research official websites, store a source-backed summary, and add a separate AI analysis for each product and ICP. Read the global role, skill and seniority vocabulary before saving a vacancy, reuse canonical identifiers, and register missing terms without case or alias duplicates. Search saved vacancies with these normalized criteria. Save public recruitment signals with sources, verification dates and vacancy status, and associate contacts with a verified career company. Find contacts or verify existing CRM emails for 1 credit per lead, with a free estimate and saved verification status, provider and date. Score the account and decision-maker against what you sell.
Keep a private, versioned my-memory.md plus one Markdown document per relationship and per company. Append selected ChatGPT, Claude, CRM or note facts with provenance for free, or confirm one estimated AI run that adds the complete grounded context in a single revision and refunds the unused reservation. Reuse a bounded Context Pack in drafting and reply triage.
List the campaigns owned by your account, configure the sender identity and inherited account signature, inspect results, or create a new campaign, attach a sender account, add or remove enrolled leads with explicit confirmation, read the authoring context, write the drafts, and repair the weak ones action by action. Find LinkedIn or Email conversations by campaign membership, specific campaign, interaction, account, status and text, with filtered totals. Find human campaign replies with scope=campaign and interaction=replied; awaiting_reply finds unanswered sends, and none includes enrolled prospects with no messages. Read incoming and outgoing LinkedIn history, inspect inbound replies, record their intent, and explicitly resume a paused prospect when the human requests it, preserving approvals and the remaining delay. Each exact email draft can hold one genuine PDF, up to 2 MB decoded, in private storage; adding or removing it always requires fresh human approval.
The inbound side on the same connection: an editorial backlog, media uploads, a publishing calendar and scheduled LinkedIn posts. Early access.
Read existing workflow runs and their progress from your agent. Web discovery workflows run in the Lead Scorer app; public research from an MCP client uses that client’s native web search, then saves verified findings in the CRM.
Lists, segments, tags, products, companies and leads. Read what is already there before adding to it, and write back everything your agent learns. Included on the free plan.
Every message the agent produces lands as a draft in your approval queue. Going live is a human action in the app — there is no tool that puts a campaign into sending on its own.
Daily cap and sending window are yours. Outreach is a slow drip you configure, not something an agent can accelerate on your behalf.
OAuth 2.1 on every supported client: you authenticate with your own Lead Scorer account and approve access from the sign-in page. Nothing to paste, nothing to rotate, and you can revoke it whenever you want.
28 short guides — the same server, framed around one client and one channel.
If you are plugging your own agent in rather than using a chat client, the agent tools page covers the workflow endpoints, the pull-and-submit pattern, and the response format. The skills library has the playbooks that drive these tools, and /llm-info is the machine-readable product summary.
https://mcp.lead-scorer.com/mcp