The MCP platform
for every team.
Manage, create, and delegate Model Context Protocol components from one place. Connect to any data source, control access, and integrate with any MCP-compatible system.
Works with the tools you trust
Platform
Everything you need
to build with MCP.
From knowledge bases to identity providers, merkur abstracts every MCP primitive into manageable components — without locking you into a single vendor.
Unified component management
Every MCP resource — tools, prompts, knowledge bases, identity providers — lives in one place, versioned and tagged for easy discovery.
Secure, granular access points
Expose exactly the components you want, to exactly the audience you define. Public or private, scoped to roles, tied to variable values.
Connect any data source
Turn GitHub repos, S3 buckets, Gmail inboxes, websites, and more into MCP knowledge bases. RAG and match queries work across containers.
Analytics and cost tracking
Monitor access point and component usage. Track token consumption and costs across your organization, all from a single dashboard.
A rich extension ecosystem
Connect to any service through built-in and custom extensions. Write in Node.js, Java, or Python. Publish extensions for your team or the community.
SDKs and CLI, built on Smithy
Official SDKs for TypeScript, Java, Python, and Go. A full-featured CLI. All generated from a Smithy model — strongly typed, versioned, and always in sync with the API.
Proxy & Integration
Wrap your existing
MCP stack.
Already have MCP resources deployed? Point merkur components at them as a proxy. Get access control, analytics, variable management, and team collaboration — without rewriting a thing.
- Layer access control on top of any MCP endpoint
- Add variable injection and secret management
- Track usage and costs across external resources
- Migrate to native components at your own pace
Extensions
Connect anything.
Extend everything.
A growing library of first-party extensions for the services your team already relies on. Or build your own in Node.js, Java, or Python.
GitHub
Repos as knowledge bases. Search code, issues, and docs.
Gmail
Email as a datasource. Search, read, and organize messages.
Tavily
Real-time web search and website ingestion for up-to-date context.
AWS
S3 buckets as file containers. Seamless cloud storage integration.
OpenAI
OpenAI models and embeddings for AI-powered components.
Anthropic
Claude model integration for reasoning and generation tasks.
Also available: Stripe, Plaid, Spotify, SMTP, SQL, Microsoft, Google, and more.
?k
Concurrent tool executions
Auto
Storage scaling
0
Config to scale
Scale
Built to scale
without thinking about it.
Tool execution scales horizontally with demand — no queues to manage, no capacity to pre-provision. Storage grows automatically as your containers ingest more data. The platform handles the infrastructure so your team handles the product.
- Tool execution scales with concurrent requests automatically
- Container storage expands as knowledge bases grow
- Access points handle traffic spikes without configuration
- Multi-tenant isolation maintained at any scale
Get started today
Your MCP infrastructure,
unified.
Join teams already managing their AI components with merkur. Free to start, built to scale.