The form backend AI agents speak natively.
splitforms ships an official Model Context Protocol server so Claude Code, Cursor, Windsurf, Cline, Continue, and Zed can read your forms, list submissions, and generate template HTML — all without you copy-pasting access keys into an IDE.
Free on every tier · No credit card · 1,000 submissions/month
What is MCP?
The Model Context Protocol is an open standard (introduced by Anthropic, now adopted by every major AI IDE) that defines a typed tool layer between an LLM and external services. Instead of an AI agent scraping API docs and guessing at request shapes, the host queries the MCP server for a schema of available tools and calls them with structured arguments.
For a form backend that means: the agent never sees your raw access key, never has to remember our endpoint URLs, and never has to parse documentation. It calls list_submissions(form_id, limit, since) the same way you'd call a local function. That's the whole pitch.
Works with every major AI coding agent.
One MCP server, every host. If your editor speaks the protocol, splitforms shows up as a tool the moment you paste in the install command.
Anthropic's official CLI — the reference MCP host. Best for end-to-end project work.
Install →AI-first IDE with first-class MCP. Great for in-editor form generation and refactors.
Install →Open-source autonomous coding agent for VS Code. Strong on long-running form scaffolds.
Install →Open-source AI assistant for VS Code and JetBrains. Lightweight inline form prompts.
Install →Install in 60 seconds.
Three steps. Same flow on every supported IDE.
sk_mcp_xxxxxxxxxxxx. Copy it now — we don't show it again.From your terminal, run our installer with the token. It writes the right config block into your IDE's mcp.json for you:
npx @splitforms/mcp install --token sk_mcp_xxxxxxxxxxxxOr paste this snippet manually into your client's MCP config:
{
"mcpServers": {
"splitforms": {
"command": "npx",
"args": ["-y", "@splitforms/mcp"],
"env": { "SPLITFORMS_TOKEN": "sk_mcp_xxxxxxxxxxxx" }
}
}
}@splitforms list submissions for my contact formTools the MCP exposes.
Seven typed tools, all read/write within the token's account, all free on every tier.
| Tool | Description | Tier |
|---|---|---|
| list_forms | List every form on the authenticated splitforms account, with id, name, and submission counts. | Free |
| get_form | Fetch a single form's schema — fields, types, validation, webhook config, allowed origins. | Free |
| list_submissions | Paginated submissions for a form, filtered by since timestamp and capped by limit. | Free |
| get_submission | Read one submission's full payload, headers, IP, source URL, and spam status. | Free |
| generate_form_html | Generate ready-to-paste HTML or JSX for a form with given fields and a styling template. | Free |
| get_access_key | Return the account's public access key so the agent can wire it into the generated form. | Free |
| create_form | Create a new form on the account with a name and a typed list of fields. Read/write scoped. | Free |
All tools are scoped to the authenticated account. Destructive operations (delete form, delete submission) are intentionally not exposed over MCP — use the dashboard.
Prompts you can paste into Claude Code today.
Six prompts that exercise the MCP end-to-end. Copy, paste, watch the agent do the rest.
Security model.
What the token can do, what it can't, and how to undo it.
Token scoping
Each MCP token is bound to the account that generated it and can only call the seven exposed tools. It cannot read billing data, change passwords, manage team seats, or touch any account other than yours. Destructive operations are not in the toolset.
Instant revocation
Open Dashboard → MCP and click 'Revoke'. Every future call from that token returns 401 immediately — no propagation delay, no caching. Generate a fresh token any time you swap machines, change agents, or suspect a leak.
Audit log
Every MCP call is logged with timestamp, IP, tool name, and arguments (sensitive fields redacted). The dashboard shows the last 30 days, exportable as CSV. You can see exactly what an AI agent did, when, and from where.
Why an MCP server for a form backend.
Three principles behind shipping AI-native tooling on every tier.
AI agents are first-class users now
In 2026 a meaningful fraction of new contact forms are written end-to-end by an AI agent. Treating those agents as second-class API consumers — making them scrape docs and copy keys — is the same mistake as ignoring mobile in 2010.
Copy-paste API keys is broken
Pasting an access key into an IDE chat means it lives in conversation history, in screen recordings, in support tickets. MCP tokens stay in a config file your agent reads but never quotes back. It's a quietly better security posture.
Generation + introspection beats docs
An agent that can call generate_form_html and immediately verify the output against a real form schema produces working code on the first try. Docs go stale. A typed MCP toolset can't.
MCP FAQ.
What is the splitforms MCP server?
It's a Model Context Protocol server that exposes splitforms as a typed toolset to AI coding agents. Instead of pasting your access key into Cursor or Claude Code and asking it to remember API docs, the agent calls structured tools like list_forms, get_submission, and generate_form_html directly. The server is free on every plan including Free.
Which AI coding agents work with the splitforms MCP?
Any MCP-compatible host: Claude Code (Anthropic's CLI), Cursor, Windsurf, Cline, Continue, and Zed are all officially supported and tested. Any other client that implements the Model Context Protocol spec works automatically — that's the whole point of the standard.
Is MCP support free or does it cost extra?
Free on every tier including the Free plan. We don't charge for AI access — submissions are still counted against your plan's monthly limit, but read/list/generate calls through MCP are unlimited and uncharged. The Pro and 4-Year plans subsidize the Free tier; AI access is a feature, not an upsell.
How do I get my MCP token?
Sign in at splitforms.com, open Dashboard → MCP, and click 'Generate token'. The token starts with sk_mcp_ and is shown once — copy it into your IDE's MCP config or run our install command. Tokens are scoped to your account only and can be rotated or revoked at any time.
Can the MCP create forms or just read them?
It's read and write. The MCP can list forms, read submissions, generate HTML, and call create_form to provision a new form on your account. All writes are scoped to the token's account — the agent cannot touch other accounts, billing, or your password. Destructive operations (delete) are intentionally not exposed.
Does MCP work with self-hosted Claude or open-weights models?
Yes — MCP is host-agnostic. If you run a local model through Ollama, LM Studio, or vLLM and pair it with an MCP-aware harness (Claude Code, Continue, Cline all support local models), the splitforms MCP works identically. No data ever leaves the path between your IDE and our API.
How is MCP different from your normal REST API?
The REST API is what you call from a browser or backend with fetch(). MCP is what an AI agent calls — it's a structured, typed tool layer on top of the same backend, with introspectable schemas the LLM uses to plan multi-step actions. You can use either or both. MCP exists so agents don't have to scrape your docs.
Can I revoke an MCP token if it leaks?
Yes, instantly. Open Dashboard → MCP, click 'Revoke' next to the token, and every future request from that token returns 401. Token usage is logged with timestamp, IP, and the tool called, so you can audit exactly what an agent did before revocation. Generate a fresh token whenever you want.