)
Connect Close MCP to your AI agents, securely.
One secured endpoint for every Close call your AI agents make, scoped per user, inspected, and logged.Try Close MCP now
What your agents can ask once Close MCP is connected.
“Leads not contacted in the past week”
“Leads with an active opportunity over $500”
“How many emails were sent this week, by user?”
“Show me the transcript of this meeting”
“Draft a follow-up email on this lead, but don't send it”
“Which opportunities are closing soonest and need attention?”
Select the tools your AI agents can call
127 tools from the Close MCP schema.
Showing 127 tools
What Close MCP exposes
Close MCP groups its 127 tools into 14 categories, each with a suggested starting policy.
| Category | Covers | Access | Suggested policy |
|---|---|---|---|
| Opportunities & Pipelines | Fetching and finding opportunities by status, owner, lead or close-date range, creating, updating and deleting them, and managing the opportunity statuses and pipelines they move through, plus listing opportunity custom fields | Read / Write | Opportunity values are the pipeline figure the business reports, so an update here changes a number people act on. A status or pipeline cannot be deleted while it is the last one or still in use, which is a useful built-in floor |
| Activities & Communications | Searching activities, fetching calls with their transcripts, individual emails, email threads with plaintext bodies, SMS messages and Notetaker meeting transcripts, reading and writing comments and notes, and creating or updating draft emails | Read / Write | The reads here are the full customer conversation record, including call and meeting transcripts and email bodies. No tool sends anything: both draft email tools state explicitly that they never send, which keeps the outbound decision with a person |
| Org & Billing | Organization and user information, memberships with roles and join dates, pending membership requests, roles with every permission they grant, groups, billing invoices, the billing summary with plan and seat counts, AI credit usage by feature, customized builtin labels, meeting outcomes and scheduling links | Read | Entirely read-only. fetch_role returns every permission a role grants and list_roles shows which grant organization management, so together they map the privilege structure. The billing tools make agent spend visible per feature |
| Custom Objects & Activities | Listing custom activity and custom object types with the fields they define, fetching and finding their instances scoped to a lead, and creating, updating and deleting those instances with their custom field values | Read / Write | The create tools instruct the agent to ask the user for field values before creating, even optional ones. On update, only the custom fields included are modified and passing null clears one, so an omission and a clear are different things |
| Templates | Fetching, finding, creating, updating and deleting email and SMS templates, and rendering an email template for a lead with its template variables resolved | Read / Write | A template is the wording that goes to customers at scale, so editing one reaches further than editing a single message. A template used in a workflow cannot be deleted, which protects live sequences |
| Leads | Fetching a lead, simple and filtered lead search, listing lead custom fields, creating, updating and deleting leads, adding and removing addresses, and merging two leads with a status check on the background merge | Read / Write | delete_lead removes the lead with all its addresses, contacts, opportunities, tasks and activities. merge_leads keeps the destination and deletes the source and cannot be undone; Close asks that the pair and the conflicts be shown and approved first |
| Tasks & Call Tasks | Fetching and finding tasks and call tasks by lead, contact, assignee, completion state and dates, creating, updating and deleting tasks, and scheduling, updating or deleting a call task assigned to a user or a voice agent | Read / Write | create_call_task sits in the destructive scope because the assignee can be a voice agent, which turns a task into a real outbound call. Completing a call task also completes every sibling sharing its deduplication key |
| Voice Agents | Listing the organization's AI callers with their configuration, objectives, instructions and enabled skills, cross-agent and per-agent performance reports, call outcomes, proposing a configuration change from natural-language feedback, applying a stored proposal, and scheduling an agent to call a lead's contact | Read / Write | propose_voice_agent_update is read-scoped and changes nothing; apply_voice_agent_update is destructive-scoped and commits the stored proposal. Gate them separately. schedule_voice_agent_call places a real phone call to a customer |
| Search & Aggregation | Natural-language search for leads and contacts referencing related activities, opportunities and tasks, fetching an arbitrary object by ID, paginating a previous search or smart view, aggregating across object types, discovering the fields aggregation can use, and searching Close's own product documentation | Read | Read-only. get_fields is the prerequisite for a valid aggregation, and search results come back with a URL that opens the same result set in Close, which is a convenient handoff to a person |
| Contacts | Fetching a contact with its names, titles, email addresses, phone numbers and URLs, listing contact custom fields, and creating, updating or permanently deleting a contact | Read / Write | delete_contact removes the person's email addresses, phone numbers and URLs from the lead and cannot be undone, though activities on the lead survive. Its description tells the agent to call it only on an explicit instruction |
| Lead Statuses | Fetching and listing the organization's lead statuses, and creating, relabelling or deleting one | Read / Write | Lead status is how the pipeline is segmented, so renaming one changes what every saved view and report means. A status cannot be deleted while leads still use it or while it is the last one |
| Smart Views | Fetching and listing lead smart views, saving a search as a smart view with its filters, sort and columns, and updating or deleting one | Read / Write | A smart view is a shared saved search colleagues work from daily, so editing one changes what a team sees without touching any record. Close asks that the name and criteria be confirmed before saving |
| Workflows & Forms | Fetching and listing workflows, creating a new workflow in Draft status, and listing the organization's web forms to resolve a form ID for a submission trigger | Read / Write | create_workflow produces a Draft rather than a live sequence, which keeps automation that contacts customers behind a human activation step. Nothing here updates or deletes an existing workflow |
| Enrichment | Using AI to determine and write the value of a field on a lead or contact from available data and external sources, returning the value with the model's reasoning and confidence | Write | It writes a model-derived value into the CRM. By default it only fills an empty field, but overwrite_existing_value replaces what a person entered, so treat that flag as the approval point |
Connect Close MCP to your AI agent
Four steps, under ten minutes.
- 01
Sign in to NeuralTrust
Open the NeuralTrust platform, sign in and go to Getting started, then Connect your tools.
- 02
Choose your agent
When asked which agent you use, pick Claude Code, Codex, Cursor, Windsurf, Copilot or your own code.
- 03
Connect Close
Find Close in the tool list and connect it.
- 04
Connect your agent
Follow the steps on screen to add the config to your agent, then send the test prompt.
Permissions down to the single tool
You do not allow Close MCP or block it. You allow the exact tools each agent can call, one by one.
| Close MCP | 127 TOOLS AVAILABLE |
|---|---|
| Code review agent | 4 tools, read only |
| Issue triage bot | 9 tools |
| Platform team assistant | all 127 tools |
- Tool by toolConnecting the server is not the unit of access. Each tool is granted on its own.
- Default denyA tool you have not enabled is never exposed to the agent.
- Approval on writesTools that change or delete things can require a human to confirm.
Is Close MCP secure?
One endpoint instead of raw credentials. Every call is checked against policy and logged.
| AI agentsConnect through your MCP client configuration. | NeuralTrustApply controls based on your configuration. | Close MCP127 documented tools organized into three nested OAuth scopes: mcp.read with 73 read-only tools, mcp.write_safe adding 17 create tools, and mcp.write_destructive adding 37 update and delete tools. Each higher scope includes all tools from the lower ones, so Close has already done the read, safe-write and destructive classification a gateway policy needs. |
|---|---|---|
| Coding assistants | Connection & credentialsper user | Activities & Communications18 tools |
| Internal copilots | Tool access policiesdeny by default | Opportunities & Pipelines15 tools |
| Workflow automations | Request & response inspectioninline | Org & Billing15 tools |
| Compatible MCP clients | Activity monitoringalerts and logs | All other tools79 tools |
Use Close MCP with any AI agent framework
Setup guides for the tools your team already uses.
- Claude CodeSetup guide
- CursorSetup guide
- CodexSetup guide
- WindsurfSetup guide
- Gemini CLISetup guide
- Google AntigravitySetup guide
- GitHub CopilotSetup guide
- Claude EnterpriseSetup guide
- Microsoft Copilot StudioSetup guide
- Open WebUISetup guide
- LangChainSetup guide
- HaystackSetup guide
- Strands AgentsSetup guide
- Vercel AI SDKSetup guide
- n8nSetup guide
Common questions about running Close MCP through TrustGate.
Other MCP servers
More servers from the catalog.
Attio
Securely connect your AI agents to Attio through TrustGate: records, lists and pipelines, notes, comments, tasks, meetings and call transcripts, emails, members, reports and read-only SQL, scoped per agent and logged.
48 toolsViewHubSpot
Securely connect your AI agents to HubSpot through TrustGate: contacts, companies, deals, tickets, engagements and associations, scoped per agent and logged.
32 toolsView)
Gorgias
Securely connect your AI agents to Gorgias through TrustGate: tickets, customers, macros, helpdesk rules, help center and guidance, AI Agent config and support actions, and analytics, scoped per agent and logged.
165 toolsView
Connect Close MCP instantly
Pick the tools, point your agent at TrustGate, and Close MCP is live: scoped, inspected, logged.