Your AI
Connect Claude, ChatGPT, Cursor, Claude Code or any MCP client to Closed Loop Enrich. It works in this workspace as you: it reads your lists, segments and qualified prospects, and anything that would spend money or write to your email platform comes back here for a person to approve.
1 · The server URL
The same address for every client, and the same connection as Closed Loop’s — one connection covers both products.
https://mcp.getclosedloop.io/mcp2 · A connect token (Cursor, Claude Code, other tools)
Claude and ChatGPT sign in with your account and need no token. For the others, issue one here; it is shown once. Manage tokens, signed-in apps and the call log on Closed Loop.
3 · Add it to your AI
Signs in with your Closed Loop account — no token to paste.
- In Claude (claude.ai or the desktop app), open Settings → Connectors and click Add custom connector.
- Name it Closed Loop and paste https://mcp.getclosedloop.io/mcp as the remote MCP server URL. Leave the OAuth client fields empty — Claude registers itself.
- Click Add, then Connect. A Closed Loop page opens: pick this workspace and click Approve.
- In a chat, turn Closed Loop on from the tools menu and send the test prompt below.
4 · Try this prompt
Use Closed Loop Enrich: list my lists, show the qualified prospects (grade A or B) on the biggest one with the reason for each, and estimate what a person-match run on its engaged readers would cost — do not start anything.
What your AI can do in Enrich
Every call is recorded with who made it. Your AI never sees an email address — it gets the domain; exports and write-backs happen here. Viewers can read but not change anything.
| Tool | What it does | When it is called |
|---|---|---|
| enrich_list_lists | Your lists with their counts, the latest run of each and any ask waiting for a person. | Reads only |
| enrich_estimate_run | What a run would do and cost, computed exactly as the run wizard does, who pays, and the projected cost per qualified lead. | Reads only |
| enrich_start_run | Starts a run. The free stages and a grade by the rubric’s rules start at once; person match, email verification and an AI grade counted against your plan become a proposal with its estimate and a hard cap. | At once if it spends nothing; otherwise a proposalApprove on Runs → Asked by your AI → Approve and start. |
| enrich_run_status | Where a run or a proposal is: stages, spend against the cap, the cost per qualified lead, warnings and errors — and a sample-first run’s sample results with the rest projected. | Reads only |
| enrich_query_contacts | Enriched contacts with grade, reason, hooks, the intent score (fit × engagement × signals, each factor explained), the verified work email and the validated phone found for A/B grades and their company’s tech, ads, hiring, news and Trustpilot reviews — the list address’s domain, never the list address. | Reads only |
| enrich_qualified_list | The qualified prospects (grade A or B) across your lists, best score first. | Reads only |
| enrich_rubrics_list | Your grading rubrics, the latest version of each. | Reads only |
| enrich_billing_overview | Your Enrich plan, the allowance used and the credits left. | Reads only |
| enrich_budget_get | This month’s Enrich spend against your monthly cap (providers and AI, your own keys included), what is left and the default Apify reserve. | Reads only |
| enrich_providers_list | The providers Enrich pays, their current prices and where each price comes from, any actor held after a price spike, the person waterfall’s order with the cost per person found, and the latest provider benchmark. | Reads only |
| enrich_benchmark_start | Asks to benchmark the person search’s providers: up to 50 people the email lookup found with confidence are searched by each later step, scored on found, right and cost, and the winner leads the order. It records the ask with its estimate and cap. | Proposal — you approve it in the appApprove on Settings → Providers → Benchmark → Approve. |
| enrich_matrix_get | A list’s engagement × fit matrix. | Reads only |
| enrich_companies_list | The companies your readers work at, with decision makers, best grade, the tech their homepage runs (email platform, CRM, ad pixels), where they run ads (Meta, Google, LinkedIn), whether they are hiring, their recent news and their Trustpilot rating. | Reads only |
| enrich_sponsor_prospects | Sponsor prospects (beta): the companies that read you and are running ads, hiring or have an ad pixel, ranked, with why — from signals already checked, nothing paid. | Reads only |
| enrich_hygiene_get | A list’s email hygiene from verification: valid, risky, invalid, bounce estimate. | Reads only |
| enrich_segments_list | Your segments, how many contacts each selects now and their offers. | Reads only |
| enrich_create_segment | Saves a segment from rules (grade, score, seniority, function, size, words, engagement, domain, verdict), read back with its count. | Changes Closed Loop only |
| enrich_draft_offers | Drafts each segment’s offer angle and opening line by Closed Loop’s rules, at no cost. Drafting with AI uses your plan’s credit, so you do that on the segment page; an offer you edited is kept. | Changes Closed Loop only |
| enrich_propose_writeback | Asks to write a list’s grades, profile fields, segment and offer — and, only when asked, the verified work address — back to the email platform it came from, as custom fields. | Proposal — you approve it in the appApprove on the list’s Write back page → Prepare, then Approve and write. |
| enrich_prospects_search | Asks to find companies that match your ideal customer (running Meta ads, in newsletter directories, on LinkedIn, or your own domain list), their decision-makers by title, and a grade of each company. It records the ask with what each step may cost, the hard cap and who pays. | Proposal — you approve it in the appApprove on Prospects → Asked for by your AI → Approve and run. |
| enrich_prospects_status | Your prospect searches: where each is, every step’s outcome with the vendor’s code and fix when one failed, what was left out and the cost against the cap. | Reads only |
| enrich_prospects_results | A prospect search’s companies, best grade first, with the reason, the public evidence and the verified decision-makers (name, title, LinkedIn) — never an address. | Reads only |
Waiting for you now: run proposals on Runs, write-back requests on each list’s page.