Agents & MCP
MCP server
The Lead Scorer MCP endpoint, how it authenticates, which clients it works with, and the full list of capabilities grouped by family.
Last updated
Lead Scorer was built AI-first: every feature ships as an MCP tool rather than as a bolted-on integration. Anything you can do in the app, an agent you control can do over the protocol — with the same credits, the same billing and the same approval gate before anything sends.
Endpoint and authentication
https://mcp.lead-scorer.com/mcp
- OAuth 2.1 is the recommended authentication, with a setup screen in the app that walks you through it.
- An Authorization: Bearer header with an MCP API key works as a manual fallback for clients without OAuth support.
- Included on every plan, free CRM included.
Supported clients
Claude (custom connectors), Claude Code, ChatGPT (developer mode), Codex, Cursor, or any MCP client you write yourself. Per-client setup instructions, with the exact commands and JSON, live on the MCP page — they are kept in one place on purpose.
Capabilities by family
The surface is grouped by capability family. Tool names are stable and are what your agent will see when it lists the server.
- CRM read
- get_lead · get_leads_from_list · get_lead_lists · get_company_info · get_companies_from_list · get_company_lists · list_tags · list_products · list_sender_accounts
- CRM write
- create_lead · update_lead · remove_lead_access · create_company · create_list · create_product · update_product · delete_product · add_leads_to_list · add_companies_to_list · add_tags_to_lead · add_lead_email · add_lead_phone · update_lead_linkedin · update_company_linkedin
- Sourcing & audiences
- search_official_company_registry · create_audience_source · list_audience_sources · sync_audience_source · fetch_profile_posts · get_lead_posts
- Enrichment & scoring
- enrich_leads · enrich_company · find_lead_contact_info · create_scoring_config · submit_lead_score · submit_lead_ai_enrichment · submit_company_classification · get_leads_pending_scoring · get_leads_pending_ai_enrichment · get_companies_pending_classification
- Campaigns & outreach
- list_sender_accounts · update_sender_account · list_campaigns · create_campaign · update_campaign_senders · get_campaign · get_campaign_authoring_context · add_leads_to_campaign · add_campaign_step · update_campaign_step · generate_campaign_drafts · write_campaign_drafts · list_campaign_actions · update_campaign_action_draft · upload_campaign_action_attachment · delete_campaign_action_attachment · invite_list_to_linkedin_event
- Content Studio
- create_content_post · update_content_post · schedule_content_post · archive_content_post · mark_content_post_published · list_content_posts · get_content_calendar · list_content_items · get_content_item · upsert_content_item · update_content_item · list_content_shows · upload_content_media
- Agent workflows
- list_workflows · start_workflow_run · advance_workflow_run · get_workflow_run · list_workflow_runs
- Personal memory & Context Packs
- get_my_memory · update_my_memory · list_relationship_memories · get_relationship_memory · list_memory_revisions · update_relationship_memory · remember_relationship_fact · archive_relationship_memory · list_memory_sources · get_memory_source · archive_memory_source · estimate_memory_source_reprocessing · start_memory_source_reprocessing · compile_context_pack · estimate_memory_ingestion · start_memory_ingestion · get_memory_ingestion_run · list_memory_proposals · review_memory_proposal
Direct Markdown reads, writes and selected sourced facts are free. Importing or reprocessing ChatGPT, Claude, transcript or note context uses Lead Scorer's AI: the estimate is reserved from that user's AI budget, actual model tokens are settled, unused budget is refunded, and each extracted fact remains a proposal until the user accepts it.
Campaign attachments are limited to one genuine PDF of at most 2,000,000 decoded bytes on one exact email draft. They stay in private storage, never belong globally to a prospect, and every upload or deletion invalidates the previous approval until a human explicitly approves that revision.
Credits over MCP
AI-powered tool calls consume the same credits and the same AI budget they would in the app. There is no cheaper back door and no separate quota to track.
Sending stays behind your approval even when an agent orchestrates the whole run. An agent can create a campaign, append a step with add_campaign_step, write every draft and queue it — it cannot launch it for you. Appended messages always require fresh human review.