Agents · MCP server

Put verification inside
your agents.

The same tool layer that powers our in-app assistant is exposed over Model Context Protocol. Point Claude Desktop — or any MCP client — at one URL and your agent inherits 40 scoped tools: verify, find, grade, clean, monitor. Scoped by key, metered by credits, confirm-first on anything billable.

One config blockNo SDK, no glue codeRevocable keys
json
// Claude Desktop · claude_desktop_config.json
{
  "mcpServers": {
    "emaillistchecker": {
      "url": "https://platform.emaillistchecker.io/mcp",
      "headers": {
        "Authorization": "Bearer elc_live_•••••"
      }
    }
  }
}

// That's it. Your agent now has 40 scoped tools —
// metered by credits, capped by your per-agent limit,
// and confirm-first on anything billable or destructive.
40
tools over one MCP server
Any
MCP-compatible client
Per-agent
credit wallets & caps
Confirm-first
on every billable call
One registry, 40 tools

Everything the platform does, as tools.

No per-tool wiring. The moment your client connects, the whole registry is available — the same functions the app itself calls.

verify_emailverify_batchgrade_listfilter_verdictsfind_emailfind_domain_emailslist_jobsjob_resultsread_analyticsexport_cleansync_integrationcheck_creditsestimate_costre_verify_unknownsinbox_placementauth_auditblacklist_checkdmarc_summarycreate_api_keylist_integrations+ more
Per-agent wallets

Give every agent a budget, not your account.

Each agent draws from your credit balance but is capped by a ceiling you set. An idle agent costs nothing — credits never expire — and a busy one physically cannot spend past its limit. Scope which tools each agent may even call.

  • Hard credit ceilings.Set a per-agent cap that can't be exceeded, ever.
  • Tool allow/deny lists. Let a bot verify and find, but never export or mint keys.
  • Fully logged. Every call is metered and auditable after the fact.
json
// Per-agent wallet — set once, enforced everywhere
{
  "agent": "outreach-bot",
  "credit_ceiling": 5000,        // hard cap; cannot exceed
  "spent_this_cycle": 1840,
  "confirm_threshold": 100,      // ask above 100 credits
  "tools_allowed": ["verify_email", "find_email", "grade_list"],
  "tools_denied":  ["export_clean", "create_api_key"]
}

// Idle agents cost nothing — credits never expire.
// A busy one physically cannot spend past the ceiling.
The guardrail travels

It asks before it spends — even in your agent.

The confirm-first guardrail that protects you in the app travels with the tools. Any billable or destructive call is proposed with its exact credit cost and waits for approval — so an autonomous agent can't quietly drain your balance or delete your data.

  • Exact cost up front. Shown in credits, before anything runs.
  • Unknown verdicts stay free.You only pay for answers we're sure of.
  • Revoke instantly. Kill a key and the agent loses access immediately.
outreach-bot wants to run
verify_batch · 4,200 addresses4,200 credits
find_email · 80 contacts80 credits
Total4,280 credits
Approve & runwithin ceiling · 5,000 cap
What you can build

Agents that act on email data.

The tools are composable — your agent chains them into real workflows.

Enrich-and-verify pipelines

Find a contact's address from name + domain, verify it, and only then write it to your CRM.

Self-healing lists

Have an agent read analytics, re-verify unknowns and export the clean set on a schedule.

Signup gatekeeping

Verify at registration and block disposables before they hit your DB.

Integration sync

Pull from Mailchimp or HubSpot, clean, and push the healthy list back.

Pre-send checks

Run inbox placement and an auth audit before an agent approves a send.

Questions

Good to know.

Model Context Protocol is the open standard for connecting AI agents to real tools and data. Point an MCP-aware client — Claude Desktop, a custom agent, an internal workflow — at our server URL and it instantly gains 40 scoped EmailListChecker tools. No SDK, no glue code.

Give your agents email superpowers — safely.

Grab an API key, paste one config block, and your MCP client has 40 tools with confirm-first guardrails. Free to start — no credit card.