Lamatic.ai MCP server
Yes. Lamatic.ai publishes an official MCP server at https://mcp.lamatic.ai/mcp. It's hosted. It offers 34 tools. We last checked it on 22 Nov 2026.
Official: answers MCP requests at mcp.lamatic.ai, on Lamatic.ai's own domain lamatic.ai. Checked 22 Nov 2026. Maintained by Stuart Brameld.
Key facts
As of 22 Nov 2026, Lamatic.ai's official MCP server scores 55/100 (provisional) on MartechDB's Agent Readiness Index (v1.1). It answered 55 of our 55 checks in the last 90 days and lists 34 tools.
How the Agent Readiness Index works · Wrong? Suggest a correction
Lamatic.ai MCP server review
55/100
Lamatic.ai's MCP server scores 55/100 on MartechDB's Agent Readiness Index: usable by AI agents, with gaps.
The good
Hosted by Lamatic.ai: connect by URL, with nothing to install
Answered 100% of our checks in the last 90 days
The bad
Nothing we've found.
Not checked yet: how you sign in; a docs page listing its tools; what agent access costs. Anything not checked scores nothing.
Official MCP server
Lamatic.ai MCP server
- Address
- https://mcp.lamatic.ai/mcp
- Runs
- Hosted
- Sign-in
- Official MCP Registry
- Not listed
- How we know it's official
-
- Answers MCP requests at mcp.lamatic.ai, on Lamatic.ai's own domain lamatic.ai.
- Tools
- 34 tools · listed 26 Oct 2026
- First seen
- 1 Oct 2026
- Last checked
-
Reachable ·
16 Nov 2026: reachable17 Nov 2026: reachable18 Nov 2026: reachable19 Nov 2026: reachable20 Nov 2026: reachable21 Nov 2026: reachable22 Nov 2026: reachable
- Uptime
- 100% · 55 of 55 checks reachable since 1 Oct 2026
Lamatic.ai MCP server tools (34)
As the server lists them. Names and descriptions are the vendor's own.
-
graph_authenticateSign in with a tokenStore a Lamatic personal access token for this connection, so the other tools can act on the user's project. Call this ONLY with a token the user has just given you. Never invent one, never reuse a token from earlier in the conversation without being asked, and never call this to "try" a value — a wrong token is rejected, not retried. If the user has no token: send them to https://studio.lamatic.ai/pat/new, where they pick the project and the permissions the token may use, then paste the result back to you. The token identifies its own project, so there is nothing else to ask them for. On success this reports the project, the role, and which tools are now usable. The token is validated before it is stored, so a failure here means the token itself is bad — relay the reason rather than retrying.
-
graph_auth_statusCheck token accessDiagnostic tool: reports which project this token owns, the role it was minted under, and exactly which tools it can and cannot use — without ever returning the token itself. Call this FIRST whenever another tool fails with an authentication or permission error, and BEFORE asking the user for any configuration: the only thing a user supplies is the personal access token itself, and everything else (project, organization, role, scopes) is read back from the token itself. The "cannot use" list is the important part — a scope a token does not carry will never work, so do not retry those tools; the user must mint a new token. Takes no arguments.
-
graph_get_guideRead a Lamatic guideRead a Lamatic reference guide. Unlike MCP resources — which many clients only expose for a user to attach — this is readable by you directly, at any point. Call it when: you are about to build or edit a flow (topic 'conventions', plus 'flow-template-chat' for a working starting point); a flow fails with an unhelpful message or a deployment succeeds without changing anything (topic 'troubleshooting'); or you need to know whether something is possible before promising it (topic 'capabilities'). Topics: 'capabilities' — scope, lifecycle, rate limits, token expiry, what is out of scope. 'conventions' — the four rules for a flow that actually runs, including the node_type vs definition_key trap. 'troubleshooting' — ranked triage for failures. 'flow-template-chat' — a verified three-node chat flow as copyable JSON.
-
graph_list_nodesList node typesList every node TYPE available for building Lamatic flows — the building blocks (triggers, LLM calls, integrations, logic, etc.), not flow instances. Each entry gives a definition key, node type, label, and builder guidance. CRITICAL — the two fields have different jobs. `node_type` is what goes in a flow node's data.nodeId; `definitionKey` is ONLY the argument graph_get_node takes. They differ for variant rows such as triggers, which are keyed <node_type>_trigger (definitionKey "chatTriggerNode_trigger", node_type "chatTriggerNode"), while standalone and response nodes have them identical ("LLMNode", "chatResponseNode"). Put the bare node_type in data.nodeId; a suffixed definitionKey there matches no executor at runtime, so the node silently pass-throughs its input and does nothing — no error, no log entry, the flow just returns the wrong thing. Whether a node is a trigger is carried by the flow node's top-level `type` ("triggerNode"), not by its nodeId. Paginated (default 50, max 20…
-
graph_get_nodeRead a node definitionGet one node TYPE's full definition JSON: its configuration fields, their types, defaults, and validation rules. This is the schema for a single node in a flow's `nodes` array — call it once per distinct node type you're about to include, before graph_create_flow or graph_update_flow, so the node's `data` object is shaped correctly instead of guessed. A 404 here means the key does not exist: re-check graph_list_nodes rather than proceeding, because a bad key produces a flow that deploys successfully and then fails at runtime.
-
graph_load_flow_detailsRead a flowLoad one specific flow's full definition: its complete node/edge graph, its trigger node and trigger type, and its input schema (parsed from the trigger's advance_schema, same as what graph_list_flows shows). Use this when you already know the flowId (from the user, or a prior graph_list_flows call) and need just that flow's details — for browsing/searching across all flows by name, use graph_list_flows instead, which returns the same schema info for every active flow in one call.
-
graph_create_flowCreate a flowCreate a NEW Lamatic flow from a node/edge graph you assemble (this cannot edit an existing flow — use graph_update_flow for that). REQUIRED procedure — do not skip steps 1-2, they prevent the two failures that silently break flows: 1. graph_list_nodes — find the nodes you need; copy each one's `definition_key` (the argument for graph_get_node) AND its `node_type` (the value that goes in data.nodeId). 2. graph_get_node for each distinct key — get its real config shape and its `node_template`. 3. graph_list_models — if any node calls a model, ASK THE USER which model and credential to use before continuing. Do not pick one yourself, even if only one credential exists. 4. Build nodes/edges, then call this tool. 5. graph_trigger_deployment — a created flow does NOT run until deployed. TWO MISTAKES THAT PRODUCE A FLOW THAT SAVES, VALIDATES, AND DEPLOYS SUCCESSFULLY, THEN MISBEHAVES AT RUNTIME: (a) Using a registry definition_key as data.nodeId instead of the bare node_type. data.nodeId mu…
-
graph_update_flowReplace a flow graphEdit an EXISTING flow in place (same flowId), unlike graph_create_flow which always makes a new one. FULL REPLACEMENT, NOT A PATCH: whatever you pass as `nodes`/`edges` becomes the flow's entire graph. Anything you omit is deleted. So for any change short of a total rewrite: (1) graph_load_flow_details to fetch the current graph, (2) edit that array, keeping everything you are not changing, (3) pass the complete result here. Passing only the nodes you touched will silently delete the rest of the flow. The same two silent-failure traps as graph_create_flow apply to any node you add or modify: (a) data.nodeId must be the bare `node_type` from graph_list_nodes ("chatTriggerNode", "LLMNode", "chatResponseNode"), NOT the registry `definition_key` with its "_trigger" / "_action_…" suffix — that key is only the argument to graph_get_node. A suffixed nodeId matches no executor and the node silently pass-throughs its input at runtime, with no error and no log entry. (b) model nodes need a real…
-
graph_execute_flowRun a flowRun a deployed Lamatic flow synchronously with an input payload and return its status and result. Before calling this, know the flow's input schema — get it from graph_list_flows or graph_load_flow_details — and build `payload` with exactly those keys; extra or missing required keys are the most common cause of a validation error or an empty/wrong result. If the flow has no input schema, pass an empty object. A non-success execution status is returned as a tool error (not a silent success) with the result payload attached, so you can see and explain why it failed. flowId is required — this server is not pinned to one flow. If this returns a bare "Workflow execution failed" with no detail, the usual causes are, in order: (1) the flow was never deployed after being created or edited — run graph_trigger_deployment; (2) a node's data.nodeId is a registry definition_key rather than the bare node_type (check against graph_list_nodes — a suffixed nodeId makes that node silently do nothing in…
-
graph_list_flowsList flowsList every ACTIVE flow in the project (inactive/draft flows are excluded), each with its description, tags (if set in Studio), trigger type, and its full input schema — the starting point for "what flows can I run" or "run the X flow" style requests, and for grouping flows by topic without guessing from the name alone. Each flow is tagged either ✅ Manually executable (safe to call directly via graph_execute_flow) or ⚠️ Not manually executable (it only fires from a real trigger event — e.g. a Cron schedule or a Google Sheets row — and calling it here with a fabricated payload will likely fail or do nothing useful; tell the user this instead of forcing it). Always call this (or graph_load_flow_details if the flowId is already known) before graph_execute_flow to confirm the flow exists and get its exact input field names.
-
graph_generate_flow_skillsGenerate flow skillsGenerate one Claude Agent Skill (a SKILL.md document) per active flow in the project — each documenting what that specific flow does, when an agent should call it, and the exact graph_execute_flow call with a matching payload. Run this ONCE per project (not per task) to make its flows individually discoverable to future sessions without repeated graph_list_flows/graph_load_flow_details calls. This RETURNS the documents to you; it does not write them. Write each to the path given above it, relative to the root of the user's project — you know where that is and this server does not, because an MCP server's working directory is wherever the client spawned it (often "/"), not the project. Then tell the user to restart this MCP client so the skills load. Pass outputDir only if the user explicitly wants the files written server-side to an absolute path they named.
-
graph_list_modelsList model credentialsList the model credentials CONFIGURED in this project, each with the credentialId, display name, and provider. This is where a real `credentialId` comes from — flow nodes that call a model (e.g. an LLM node's generativeModelName[].credentialId) must reference one of these IDs. ALWAYS call this before building or editing a model-using flow with graph_create_flow / graph_update_flow, and never invent or placeholder a credentialId: a flow saved with a fake one passes validation but fails at runtime with a generic "Workflow execution failed". If this returns an empty list, the project has no model set up yet — use graph_list_model_providers and graph_create_model_credential first. ASK THE USER WHICH MODEL AND CREDENTIAL TO USE — do not choose for them. The choice affects cost, quality, latency and which vendor sees their data, so present what this returns and let them pick, unless they have already named both. Having exactly one credential is not permission to skip asking: one credential…
-
graph_list_model_providersList model providersList the model providers the Lamatic platform SUPPORTS (e.g. openai, google, anthropic) — not what this project has configured; for that use graph_list_models. Each provider lists the credential fields it requires, which is exactly what graph_create_model_credential's `credentials` object must contain. Set includeModels to also get each provider's available model slugs (the values a flow node's model_name expects, and what graph_get_model_status checks).
-
graph_get_model_statusCheck a modelCheck whether a specific model is currently available on the platform. The argument is named `modelName` and takes the model SLUG (e.g. "gemini-1.5-pro", "gpt-4o") — not a credential id and not a provider. Use this to verify a model before wiring it into a flow, or to check whether an unavailable model explains a failing flow. Get valid slugs from graph_list_model_providers with includeModels set to true.
-
graph_create_model_credentialAdd a model credentialAdd a model provider credential (e.g. an API key) to the project, returning a new credentialId that flow nodes can reference. Call graph_list_model_providers FIRST to learn exactly which credential fields the chosen provider requires, and pass those as the `credentials` object. Credential names must be unique — reusing an existing name is rejected. SECURITY: the secret you pass here travels through this conversation; prefer having the user add credentials in Lamatic Studio directly, and only use this tool when they have explicitly asked you to and supplied the key for that purpose.
-
graph_list_deploymentsList deploymentsList the project's deployments, most useful for checking whether recent flow changes have actually been pushed live and whether the last deployment succeeded. Returns each deployment's id, name, and status (e.g. deploying / success / failed). Use graph_get_deployment for the full detail of any one of them. Note that flow edits made with graph_create_flow / graph_update_flow are NOT live until a deployment runs — see graph_trigger_deployment.
-
graph_get_deploymentRead a deploymentGet one deployment's full record: name, description, status, the list of changes it deployed, who/when it started, and how long it took. Use this to confirm a deployment finished successfully, see exactly which changes went live, or investigate a failed or slow deploy. Get deployment IDs from graph_list_deployments or from the result of graph_trigger_deployment.
-
graph_trigger_deploymentDeploy the projectStart a new deployment, pushing flow definitions live. This is the step that makes changes from graph_create_flow / graph_update_flow actually take effect — editing a flow alone does not deploy it, and the flow will not execute until a deployment that INCLUDES it has succeeded. By default this deploys every flow in the project; pass flowIds only to deploy a specific subset. Deployment is asynchronous: this returns once the deploy is kicked off (usually status 'deploying'), so poll graph_get_deployment with the returned ID to confirm it actually succeeded. Confirm with the user before deploying — this changes what is running in their live project.
-
graph_get_logsRead execution logsFetch paginated execution logs for a project — one entry per flow run, with timestamp, flow name, execution type, response status, duration, and cost. Use this to check whether a flow actually ran, debug why an execution failed or was slow, or answer "how much has this cost" style questions. Logging must be enabled for the project (graph_enable_logs) for new entries to appear; if this returns nothing and you expect runs to exist, check that first with graph_auth_status or by trying graph_enable_logs.
-
graph_d1_list_tablesList tablesList every table in the project's D1 (SQLite) database, with each table's column count and row count. Call this first to discover what tables exist and roughly how big they are, before reading columns (graph_d1_get_table_columns) or querying/altering data (graph_d1_execute_query) — table names are case-sensitive and must match exactly.
-
graph_d1_create_tableCreate a tableCreate a new table in the project's D1 (SQLite) database from a column definition. This only defines the schema — the table starts empty; insert rows afterward with graph_d1_execute_query. Fails if a table with this name already exists — check graph_d1_list_tables first if you're unsure.
-
graph_d1_drop_tableDrop a tablePermanently delete a table and ALL of its rows from the project's D1 database. This is destructive and cannot be undone — confirm with the user before calling this, especially for any table that might hold data worth keeping. Use graph_d1_get_table_columns or graph_d1_execute_query to inspect a table's contents first if you're unsure it's safe to drop.
-
graph_d1_get_table_columnsRead table columnsGet the column definitions (name, type, primary-key flag, nullability, default value) for one D1 table. Use this before writing an INSERT/UPDATE query against a table you haven't seen the shape of, or to confirm a graph_d1_create_table call took effect as expected.
-
graph_d1_execute_queryRun SQLRun a raw SQL statement (SELECT, INSERT, UPDATE, DELETE, or DDL) against the project's D1 (SQLite) database and get back the resulting rows plus query metadata. There is no dry-run, transaction, or rollback — a write statement takes effect immediately. Check graph_d1_list_tables / graph_d1_get_table_columns first if you don't already know the table's shape, so the query references real columns.
-
graph_disable_logsTurn logging offTurn off execution logging for the project going forward. Existing log entries in graph_get_logs are kept — this only stops new ones from being recorded. The change requires the project to be redeployed before it takes effect, so logs may keep appearing for a bit after this call; tell the user that if it comes up.
-
graph_enable_logsTurn logging onTurn on execution logging for the project going forward, so future flow runs start showing up in graph_get_logs. The change requires the project to be redeployed before it takes effect, so don't expect graph_get_logs to show a run made immediately after this call.
-
graph_vectordb_create_collectionCreate a collectionCreate a new VectorDB collection ("store") in the project — either for semantic search over embeddings (`type: "vector"`, e.g. a RAG knowledge base) or for agent memory (`type: "memory"`). Confirm with graph_vectordb_list_collections afterward that it exists before referencing it elsewhere. Note: there is no dedicated "add record" tool — records are typically populated by the project's own ingestion flows; graph_vectordb_update_record only modifies a record that already exists.
-
graph_vectordb_get_collectionRead a collectionGet one VectorDB collection's details: its type, provider, schema properties, and how many objects it currently holds. Use this to check a collection's shape before writing a query or record against it, or to confirm graph_vectordb_create_collection succeeded.
-
graph_vectordb_list_collectionsList collectionsList every VectorDB collection ('store') in the project, with each one's type, provider, and object count. Call this first to discover what collections exist and find a collectionId, before reading/listing records or getting a single collection's details.
-
graph_vectordb_list_recordsList recordsList records stored in a VectorDB collection, paginated. This is a plain listing (no semantic/similarity search) — useful for browsing what's in a collection or finding a specific record's ID before calling graph_vectordb_get_record or graph_vectordb_update_record. If you need the collectionId first, call graph_vectordb_list_collections.
-
graph_vectordb_get_recordRead a recordFetch one record by its exact ID. Only needed when the recordId came from OUTSIDE a listing — a flow output, a log line, a user pasting it — or to re-read a record after graph_vectordb_update_record. Do NOT chain this after graph_vectordb_list_records or graph_vectordb_search: both already return each record's full contents, so calling this again returns nothing new.
-
graph_vectordb_searchSearch recordsSearch a VectorDB collection and get records back RANKED by relevance, each with a score. This is the tool to use for any 'find records about X' question — graph_vectordb_list_records only browses in storage order and cannot answer it. Pick searchType deliberately: - 'hybrid' (default) combines meaning and exact wording. Best for most questions. - 'vector' matches by meaning only. Use when wording will differ from the stored text. - 'keyword' matches exact words (BM25). Use for names, codes, error strings — and note it is the ONLY type that needs no embedding model, so it works in projects with no model credential. `model` and `credentialId` are only for 'vector' and 'hybrid'. Leave both out first: they are inferred when the project has exactly one model credential and that provider offers exactly one embedding model. If either is ambiguous the error lists the valid values — pass one of those back rather than guessing. Do not invent a model name; ask the user which to use if the list…
-
graph_vectordb_delete_collectionDelete a collectionPermanently delete a VectorDB collection and ALL of its records. This is destructive and cannot be undone — confirm with the user before calling this on a collection that might hold data worth keeping (e.g. a populated RAG knowledge base). Use graph_vectordb_get_collection first to check its object count if you're unsure.
-
graph_vectordb_update_recordOverwrite a recordUpdate fields on a record that already exists in a VectorDB collection (this does NOT create new records — the recordId must already exist, e.g. one found via graph_vectordb_list_records). Use graph_vectordb_get_collection to see the collection's property schema first, so `data` uses the right field names and types.
Connect your AI to Lamatic.ai
Add the official Lamatic.ai MCP server to Claude, ChatGPT or your editor.
Connect Lamatic.ai MCP to Claude
Claude.ai and Claude Desktop
Open Customize, then Connectors.
Click +, then Add custom connector.
Name it Lamatic.ai, paste the address below and click Add.
Connect Lamatic.ai MCP to ChatGPT
Open Settings, then Apps, then Advanced settings.
Turn on Developer mode, then click Create.
Name it Lamatic.ai and paste the address below.
ChatGPT usually only connects to servers with self-serve sign-in or none.
Connect Lamatic.ai MCP to Cursor
Opens Cursor if it's installed. If nothing happens, use the JSON below.
Connect Lamatic.ai MCP to VS Code
Opens VS Code if it's installed. Otherwise add the JSON settings from the Cursor tab to your .vscode/mcp.json.
Lamatic.ai MCP server questions
Does Lamatic.ai have an official MCP server?
Yes. Lamatic.ai publishes its MCP server itself. Its address is on Lamatic.ai's own domain (mcp.lamatic.ai). MartechDB only lists servers a vendor publishes, never community copies.
What is the Lamatic.ai MCP server URL?
The address is https://mcp.lamatic.ai/mcp.
How do I connect Lamatic.ai MCP to Claude, ChatGPT or Cursor?
In Claude, add https://mcp.lamatic.ai/mcp as a custom connector (Customize, Connectors, Add custom connector). In Cursor, use the Add to Cursor link on MartechDB's page, or paste the JSON settings into your mcp.json. In Claude Code, run: claude mcp add --transport http lamaticai https://mcp.lamatic.ai/mcp
What can the Lamatic.ai MCP server do?
It offers 34 tools, including graph_authenticate, graph_auth_status, graph_get_guide, graph_list_nodes and graph_get_node (as listed on 26 Oct 2026). 19 are marked read-only and 5 can delete or overwrite data.
Is the Lamatic.ai MCP server in the official MCP Registry?
No. Lamatic.ai hasn't published it to the official MCP Registry.
Is the Lamatic.ai MCP server working?
Yes. It answered our most recent check. We last checked it on 22 Nov 2026.
What are the alternatives to Lamatic.ai's MCP server?
In Data integration, the most agent-friendly alternatives with an official MCP server are Needle (97/100), Supermetrics (97/100), Activepieces (97/100), IFTTT (92/100) and import.io (92/100). Lamatic.ai scores 55/100 on the same index.
Alternatives to Lamatic.ai's MCP server
Vendors in Data integration with an official MCP server, the most agent-friendly first. Lamatic.ai scores 55/100.
Needle
needle.app
Supermetrics
supermetrics.com
Activepieces
activepieces.com
IFTTT
ifttt.com
import.io
import.io
Use with AI
Connect your AI to MartechDB
Then ask it about Lamatic.ai, or any of the 1,850 marketing tools with an official MCP server. Free, no sign-in.
Claude.ai and Claude Desktop
Open Customize, then Connectors.
Click +, then Add custom connector.
Name it MartechDB, paste the address below and click Add. There's no sign-in.
The free plan allows one custom connector. On Team and Enterprise plans, an owner adds the connector first.
Claude Code
Run this in your terminal:
ChatGPT
Open Settings, then Apps, then Advanced settings.
Turn on Developer mode, then click Create.
Name it MartechDB, paste the address below and choose No authentication.
Developer mode is on Plus and Pro plans, and on Business, Enterprise and Edu if an admin allows it.
API
The same data as JSON, with no sign-in. See the OpenAPI description for every endpoint and field.
CLI
Search the directory and print the command to connect a vendor's server, from your terminal. Needs Node 18 or later. View on npm
Show that your MCP server is verified
If you work at Lamatic.ai, add this badge to your MCP docs or README. It's free, it's the same for every vendor, and it updates from our daily checks: it shows uptime while the server responds, and says so if it stops.
Something wrong or out of date? Suggest a correction or email us, and we'll check it against the evidence. Every correction we make is listed on Corrections.