Claude (web, desktop, phone)
Add a custom connector with the server address, then sign in.
MCP server for final expense agencies
Run your book from the AI you already use
Connect Claude, ChatGPT, Grok, Cursor or any MCP app to LegacyCore. 104 tools for policies, drafts, at-risk policies, commissions, carrier status and Do-Not-Call checks, scoped to the book you are allowed to see. Included on every plan.
Server address
Connect in about a minute
Every app uses the same address. You sign in with your LegacyCore account on a LegacyCore page and choose how much the app may do. Nothing else to paste.
ChatGPT
Add a custom connector (MCP) with the server address, then sign in.
Grok
Add a connector with the server address, then sign in.
Sesame
Apps › Add custom app: name it LegacyCore, paste the server address, then Continue to authorization.
Cursor
Add an MCP server with the server address as its url, then choose Login when Cursor asks.
Claude Code
Add the server, then run /mcp and choose LegacyCore to sign in.
Codex CLI
Add the server, then sign in from the terminal.
Unattended agents
Nobody is there to approve a sign-in, so a scheduled or background agent uses a scoped LegacyCore API key, sent as Authorization: Bearer lc_…. Create it in Settings › API keys.
Just ask
Questions agents and owners ask once LegacyCore is connected.
- “Which of my policies missed a draft this month?”
- “Start conservation on the at-risk ones over $50 a month.”
- “What’s my booked advance, and how much is still pending?”
- “Where is the Americo application I sent this morning?”
- “Turn the Johnsons’ beneficiaries into referrals.”
- “Is this number on the Do-Not-Call list?”
Ready-made workflows
Your AI app lists these as prompts once LegacyCore is connected. Set the weekly retention check as a scheduled task in Grok or Claude and it lands every Monday morning.
Daily book review
Money, missed premiums, at-risk policies, stuck submissions and today's calendar in one brief
Chase missed payments
Find policies with a missed or NSF draft and line up the save: DNC check, conservation, a callback
Weekly retention check
Made to run as a weekly scheduled task: first drafts not yet paid, missed or NSF drafts inside the advance window, and the save list
New sale follow-up
After a policy is written: send the policy summary, harvest referrals, book a 30-day check-in
Portal worker triage
Walk every stuck carrier submission: what happened, whether to retry, and what the agent must do
Agent performance review
Production, money, call quality and goals for one agent, with three coaching points
What it can do
A selection of the 104 tools, by area. The full list, with what each one reads or changes, is in the public server card.
Your book
Find a policy, read its history and notes, update its status, or add a note.
- list_applications
- get_application
- get_application_timeline
- get_application_notes
- update_application_status
- update_application
- append_application_note
- resolve_case
Drafts and retention
See each month’s premium draft, find the policies at risk, and start conservation follow-up.
- get_policy_payments
- update_policy_payment
- list_failed_payments
- list_at_risk_policies
- trigger_conservation
- get_conservation_queue
Commissions
Booked advance, paid, pending and back end due, the same figures as the Finance page.
- list_commissions
- get_revenue_summary
- get_earnings_statement
- update_commission_status
- get_goals
Carriers and submission
Pull carrier status into the book, review policies the carrier has that you don’t, and follow Portal Worker application jobs.
- carrier_sync_list_recent_runs
- carrier_sync_dry_run
- carrier_sync_create_missing_applications
- portal_worker_status
- portal_worker_list_jobs
- portal_worker_get_job_progress
- portal_worker_recover_job
Calls and coaching
Transcripts and summaries of recorded calls, live call sessions and coaching insights.
- list_voice_calls
- get_call_transcript
- get_call_summary
- list_live_call_sessions
- get_coaching_insights
- get_agent_performance
Compliance
Check a number before outreach, honor a stop request, and read the compliance log.
- check_dnc_status
- add_to_dnc
- set_application_dnc_status
- get_dnc_audit
- get_compliance_log
Referrals and follow-up
Turn an active policy’s beneficiaries into referrals, book a follow-up, or send the client their policy summary.
- harvest_referrals
- list_referrals
- list_funnel_leads
- create_follow_up_event
- get_calendar_events
- send_policy_summary
- list_letters
Running the agency
Agents, team performance, the daily report and system health, for managers and admins.
- list_agents
- get_analytics_report
- get_daily_report
- list_access_sessions
- get_system_health
You stay in control
Read only
See your book, commissions, calls and reports. Cannot change anything.
Read and act
Everything in read only, plus update policies, payments, notes, DNC and follow-ups in your book.
Full access
Every tool across the whole agency, including admin controls. Admins only.
- Agents reach only their own book; managers reach their downline.
- Every tool call is logged with the tool name and the connection that made it.
- Conservation SI follow-up waits for a person to approve it before it sends.
- The Portal Worker stops before the carrier’s final Submit, so nothing is filed until an agent approves it.
Questions
Yes. LegacyCore runs a remote MCP (Model Context Protocol) server at https://www.legacycore.io/api/mcp, so Claude, ChatGPT, Grok, Cursor and other MCP apps can work with your final expense book. You sign in with your LegacyCore account and choose read only or read and act; full access is for admins. It has more than 100 tools covering policies, premium drafts and missed payments, at-risk policies and conservation, commissions, carrier status sync, Portal Worker jobs, call transcripts and Do-Not-Call checks, and it comes with every plan. The public tool list is at https://www.legacycore.io/.well-known/mcp/server-card.json.
Claude (web, desktop and phone), ChatGPT, Grok, Sesame, Cursor, Claude Code and Codex CLI sign in with your LegacyCore account using the one server address, https://www.legacycore.io/api/mcp. Other MCP apps that support Streamable HTTP with OAuth sign-in connect the same way. Unattended agents, with nobody there to approve a sign-in, use a scoped LegacyCore API key instead.
Only if you grant it Read and act. Then it can update policy status, premium payments, notes, Do-Not-Call flags and follow-ups in your own book; with Read only it cannot change anything. Agents reach only their own book and managers their downline. Every tool call is logged with the tool name and the connection that made it. Conservation SI follow-up still waits for a person to approve it before it sends, and the Portal Worker still stops before the carrier's final Submit.