# Lead Scorer documentation

> Public documentation for Lead Scorer, an AI SDR for solo founders. Agents find real companies, score them against what you sell, write the LinkedIn and email messages and queue them; nothing sends without human approval. The whole product is exposed as an MCP server (OAuth 2.1), so any agent can drive it. Every page below is public and needs no account.

## Start here

What the product is and what happens the first time you run it.

- [Getting started](https://lead-scorer.com/docs/getting-started): What Lead Scorer is, what happens during your first run, and the two ways to drive it — the app or your own AI agent.
  - Sections: Two ways to drive it · What a first run looks like · What you need before you start · Where to go next

## Concepts

The handful of objects everything else is built on.

- [Core concepts](https://lead-scorer.com/docs/core-concepts): Personal memory, products, ICP, scoring and credits — the objects the rest of Lead Scorer is built on.
  - Sections: Personal and relationship memory · Product · ICP · Two-level scoring · Campaigns · Credits and AI budget
- [Agent workflows](https://lead-scorer.com/docs/agent-workflows): How a run works: the phases it goes through, where you are asked to approve, how to replay it, and how the daily drip works.
  - Sections: The phases of a run · Replaying a run · The daily drip · Running workflows from your own agent

## Agents & MCP

Driving Lead Scorer from Claude, ChatGPT, Cursor or your own agent.

- [MCP server](https://lead-scorer.com/docs/mcp): The Lead Scorer MCP endpoint, how it authenticates, which clients it works with, and the full list of capabilities grouped by family.
  - Sections: Endpoint and authentication · Supported clients · Capabilities by family · Credits over MCP
- [Agent skills](https://lead-scorer.com/docs/skills): The downloadable playbooks that tell your agent how to use the Lead Scorer tools — categories, learning path and how to install them.
  - Sections: Installing a skill · What the library covers · The learning path

## Related resources

- [Changelog](https://lead-scorer.com/changelog): what shipped to production and when, in user language. Latest entry: 2026-08-06.
- [Agent skills library](https://lead-scorer.com/skills): 24 downloadable SKILL.md playbooks covering 55 MCP tools.
- [Skills archive](https://lead-scorer.com/skills/download): the whole library as a single download.
- [MCP setup](https://lead-scorer.com/mcp-server): per-client configuration for Claude, Claude Code, ChatGPT and Cursor.
- Agent-native setup: OAuth can create the account during MCP connection; the agent can then read readiness, sell only Pro monthly or top-ups, poll payment, connect email/LinkedIn, and complete UI onboarding.
- [Product overview for AI assistants](https://lead-scorer.com/llms.txt): the site-wide llms.txt, covering positioning, pricing and comparisons.

## Notes

- Company discovery runs on the official French State registry (recherche-entreprises.api.gouv.fr, built on SIRENE and the INPI RNE). Every account carries a verifiable SIREN.
- Every outreach draft follows a versioned playbook; placeholders and email HTML are sanitized before it reaches the human review queue.
- There is no auto-send: sending always requires an explicit human approval, including for runs orchestrated over MCP.

Documentation last updated: 2026-08-20