Self Hosted
Deploy doc2mcp inside your own VPC. Your docs and your generated MCP servers never leave your perimeter.
Any URL to Model Context Protocol
Turn Any DocumentationInto AI Tools
Into AI Tools
Paste a docs URL—Mintlify, Docusaurus, GitHub, OpenAPI—and get a hosted, Cursor-ready MCP server in seconds. No setup, no local scripts.
Try an example:
<60s
docs → MCP
PIPELINE
1 URL
paste & go
INPUT
Hosted
no install
REMOTE
5+
MCP clients
EXPORTS
Ecosystem
Built for the modern AI ecosystem
One MCP server, ready to plug into every major AI editor and agent runtime.
- Cursor
- Claude Desktop
- VS Code
- Claude Code
- Windsurf
- Codex
- Gemini CLI
- ChatGPT
- Zed
Plus any MCP-compatible AI tool via manual configuration.
Platform
Oneplatform.Everylayer.
From crawl to retrieval to deployment — the full stack between your docs and an AI agent.
6 Capabilities1 Secure Endpoint
Smart Documentation Crawling
Automatically understands documentation hierarchy and structure — Mintlify, Docusaurus, Swagger, GitBook, raw markdown.
Target Tree● Scan complete
stripe-docs
├── api-reference
│ ├── charges.mdx12kb
│ ├── customers.mdx18kb
│ └── refunds.mdx9kb
├── checkout
└── billing
Metadata & Statsdoc2mcp-crawler/1.0
Host URLdocs.stripe.com
Total Pages1,284 pages
Data Scanned12.4 MB
Crawl Speed52 pages/sec
Security checks:100% Passed
Production Ready MCP Generation
Generate MCP servers without writing MCP code. Tools, workflows, auth, hosting — done.
Generated Tool Schema
// stripe/create_payment_intent
type Input = {
amount: number;
currency: string;
metadata?: Record<string, string>;
};
Agent Optimized Retrieval
Improve AI coding accuracy and reduce hallucinations with semantic, schema-aware retrieval.
Query:stripe session return URL parameters
Retrieve Payment Intentpayment_intents.mdx
94% matchCreate Checkout Sessioncheckout_sessions.mdx
89% matchLive Documentation Sync
Keep documentation synchronized automatically — your MCP server tracks the source.
docs.stripe.com2 mins ago
12 pages updated
stripe-openapi.json1 hr ago
1 tool generated
Multi-Agent Compatibility
Works across the modern MCP ecosystem — Cursor, Claude, VS Code, Windsurf, OpenAI Agents.
One Click Export
Instant MCP configuration generation. Copy a JSON snippet, paste into your editor, done.
"stripe-mcp": {
"command": "npx",
"args": [
"-y",
"doc2mcp-server@latest",
"--key=st_3a1"
]
}
<TESTIMONIALS>
Developer Love
Outcomes
Why teams use Doc2MCP
The compound effect of giving every agent — internal or customer- facing — structured, current, AI-ready documentation.
Reduce hallucinated code
Agents respond from your real docs — not stale training data — so generated code actually compiles and ships.
Improve AI coding accuracy
Schema-aware retrieval narrows answers to the right endpoint, the right version, the right example.
Avoid manual MCP development
Stop hand-writing tool wrappers per product. Doc2MCP generates them from documentation, automatically.
Keep documentation updated
Live sync keeps the MCP server tracking your source — without redeploys, manual diffs, or token rotation.
Give AI agents structured context
Replace prompt stuffing with structured, retrievable, AI-optimized context — usable across every MCP client.
Ship integrations faster
From “read docs” to “write integration” in minutes — across Cursor, Claude, Windsurf, OpenAI Agents.
Scale internal knowledge systems
Convert internal runbooks, ADRs, RFCs, and product docs into a queryable knowledge layer for your agents.
Reclaim engineering hours
Stop pasting docs into prompts and debugging hallucinations — your team’s time goes back to product.
Use cases
Infrastructure for every AI-native team.
From public APIs to private wikis — doc2mcp converts any documentation surface into a hosted MCP server your agents can actually use.
Live Deployment TargetActive
stripe-mcp-server● Public Registry
Available Tools (23):
· create_charge
· retrieve_customer
· list_payment_intents
Key Features
- OpenAPI, Swagger, and Postman collections
- Auto-generated auth + workflow primitives
- One MCP URL, distributed to every customer's IDE
SOC 2 ready architecture
AI-native indexing
Agent-shaped toolkits
< 60s URL → MCP
Comparison
Manual MCP development vs Doc2MCP
Build it by hand and you ship a one-off, brittle integration. Use Doc2MCP and you ship infrastructure — structured, current, and ready for every agent in your stack.
Capability
Manual MCP development
DIY · slow · brittle
Doc2MCP
Infrastructure · automated · current
Setup time
Days to weeks of engineering
Minutes from a docs URL
Maintenance
Manual diffs on every API change
Continuous, automated sync
Documentation understanding
You read it, model guesses
Structured, schema-aware context
Updates
Redeploy, regenerate tools
Live — same endpoint, fresh content
Retrieval layer
DIY embeddings + glue code
Semantic + workflow-aware out of the box
AI readiness
Endpoint-shaped tools
LLM-shaped tools and workflows
Deployment
Self-host a server per integration
One hosted remote MCP, multi-tenant
Support
Internal, ad hoc
Dedicated team, audit logs, SLAs
EnterpriseEnterprise-ready
Enterprise-ready
AI infrastructure.
The same MCP layer your engineers love — hardened with the controls, audit trails, and deployment options enterprise teams require.
Private Deployments
Single-tenant infrastructure on AWS, GCP, or Azure with bring-your-own-key encryption for every artifact.
Access Control
SSO via SAML / OIDC, RBAC for projects + workspaces, and short-lived MCP tokens scoped to a specific team.
Audit Logs
Every MCP query is logged with actor, tool, and document context — exportable to Datadog, Splunk, or S3.
Team Workspaces
Organize projects by team, share MCP servers across squads, and bill at the org level — no per-seat surprises.
Custom Pipelines
Bring your own crawler, embedder, or post-processing step via webhook — doc2mcp orchestrates the rest.
Usage Analytics
See which tools your agents actually call, which docs are hot, and where retrieval quality is degrading.
Enterprise Security
SOC 2-ready controls, signed MCP token rotation, prompt-injection guards, and per-project secret isolation.
SOC 2 in progress
GDPR-ready data handling
EU + US regions
Why now
Why this matters right now.
AI agents are becoming the next application layer. MCP is emerging as the standard protocol for tool access. Millions of APIs and documentation sites are not MCP-ready — doc2mcp bridges that gap automatically.
“Documentation is becoming the knowledge layer for AI agents.”
67%
Agents are the new interface.
Cursor, Claude, and Windsurf now write the majority of new code at AI-native startups. The interface to software is no longer a UI — it's an agent.
of new code at AI-native teams ships via agents
5,000+
MCP is becoming the standard.
Anthropic, OpenAI, and the broader ecosystem have converged on the Model Context Protocol as the canonical way agents discover and use tools.
servers listed in the official MCP Registry
< 1%
Most docs are not MCP-ready.
Millions of API references, SDKs, and developer portals were written for humans. There is no hosted, semantic, agent-ready layer in front of them.
of public docs expose a production MCP
< 60s
doc2mcp closes the gap.
We turn any docs URL into a hosted, agent-ready MCP server in under a minute — no manual coding, no infrastructure, no per-vendor lock-in.
from URL paste to a Cursor-ready MCP
Stats reflect internal observations across AI-native teams · May 2026
CLI · available nowThe whole pipeline,
The whole pipeline,
from your terminal.
Install the doc2mcp CLI to convert docs, generate MCPs, and wire them into Cursor, VS Code, Claude, and Windsurf — all from your terminal with the same limits as the web app.
- One command from docs URL to live MCP
- Connect editors without leaving the terminal
- Run conversions and syncs straight from CI
Available on npm
doc2mcp login · doc2mcp <docs-url>
doc2mcp — zsh
$ n
PRICING PLANS
Documentation infrastructure for AI agents.
For developers who want to generate and self-host MCPs — and for companies publishing official MCP infrastructure for their customers and AI agents. Token-based access, hosted endpoints, and custom domains included.
Free
For trying doc2mcp and one-off experiments.
₹0/ forever
- 1 MCP conversion / month
- Up to 30 pages per docs site
- Token-based remote MCP (Cursor + Claude)
- ask_documentation with AI
- Public projects · community support
Starter
For individual developers self-hosting MCPs.
₹299/mo
Charged once as ₹299
- 5 MCP conversions / month
- Up to 50 pages per docs site
- Export MCP config · self-host anywhere
- Token-based HTTP MCP (Cursor + Claude + Windsurf)
- Semantic toolkits + workflow inference
- Public projects
Most popular
Pro
For startups & API companies shipping official MCP.
₹999/mo
Charged once as ₹999
- 10 MCP conversions / month
- Up to 750 pages per docs site
- Hosted MCP · auto re-crawl every 24h
- Custom domain for your MCP endpoint
- Workflow AI + visual API graph
- Usage analytics · private projects
- Email support
Team
For companies operating official MCP infrastructure.
₹2,999/mo
Charged once as ₹2,999
- Everything in Pro
- Up to 2,500 pages per docs site
- Custom domain + company-attributed tokens
- Auto re-crawl every 6h
- 5 teammates · role-based access
- Priority support (24h SLA)
Unlimited MCP reads included · Conversions run once per docs portal · Cancel anytime.
FAQ
Questions, answered.
Everything developers and teams ask before generating MCP servers.
The Model Context Protocol is an open standard from Anthropic that lets AI agents (Cursor, Claude Desktop, Windsurf, OpenAI Agents) discover and call tools over a uniform JSON-RPC interface. doc2mcp produces fully compliant MCP servers your agents can connect to with a URL + Bearer token.
Paste a documentation URL. We crawl the site (Mintlify, Docusaurus, Swagger, GitBook, plain HTML — all supported), preserve code blocks, run an AI understanding pass that detects auth, workflows, and endpoint groups, and generate a hosted MCP server with semantic tools — typically in under 60 seconds.
Anything that speaks MCP — Cursor, Claude Desktop, VS Code, Windsurf, OpenAI Agents, Cline, Zed, and the broader MCP client ecosystem. We provide ready-to-paste config snippets for every major client in your project dashboard.
Yes. Pro and Team plans support private projects with authenticated crawling and per-MCP Bearer tokens. Enterprise customers can self-host doc2mcp inside their VPC with SSO, RBAC, and audit logging — your docs and tokens never leave your perimeter.
Free plans run on-demand refreshes. Pro plans auto-resync every 24h, Team plans every 6h, and Enterprise plans support webhook-triggered instant updates. We use content-hash diffing so only changed pages get re-crawled.
Yes. We offer fully self-hosted deployments on AWS, GCP, or Azure with bring-your-own-key encryption, SSO via SAML/OIDC, audit logs, and custom retention policies. Reach out via the contact form to start a security review.
Still curious? Talk to the founders →
Built for the people building the agentic web.
5+
MCP clients supported
v0.1
Live in the MCP Registry
< 60s
Docs to a live MCP