The omnistream-mcp server connects AI assistants to the OmniStream marketplace, enabling discovery and direct interaction with 195+ REST APIs and real-time data streams through a single MCP integration.
Browse APIs (list_apis): List all published APIs, optionally filtered by category, with details like slug, title, rating, uptime, and endpoint count.
Search APIs (search_apis): Perform BM25 keyword search across API titles, descriptions, tags, and endpoint names to find relevant APIs by topic.
Get API details (get_api): Retrieve full details for a specific API including description, base URL, authentication method, and all endpoints with parameters.
List endpoints (list_endpoints): Get a lightweight list of all endpoints for a specific API (operation IDs, HTTP methods, paths, required parameters) without the full API detail payload.
Call any API (call_api): Proxy-call any endpoint on any published API using your Omni key — supports query parameters and JSON request bodies, with upstream credentials injected server-side and never exposed to the AI.
Browse real-time streams (list_streams): List all available real-time data streams on the OmniStream Grid, including stream ID, category, update interval, and subscriber count.
Inspect a stream (get_stream): Get details for a specific real-time stream, including its update interval and the last 5 data samples to understand the payload structure.
Note: Only
call_apirequires theOMNI_KEYenvironment variable; all other tools are keyless.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@omnistream-mcp What APIs are available on OmniStream?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP (Model Context Protocol) server for OmniStream. Lets AI assistants discover and call any marketplace API directly.
npx omnistream-mcpExposes the OmniStream marketplace as MCP tools that AI assistants (Claude Desktop, Cursor, Continue, etc.) can call:
Tool | Auth needed | Description |
| No | Browse all published APIs |
| No | BM25 keyword search |
| No | Full API details + all endpoints |
| No | Endpoint list for one API |
| Yes (OMNI_KEY) | Proxy-call any API endpoint |
Related MCP server: decixa-mcp
Most MCP servers wrap a single service, so covering ten APIs means installing and configuring ten servers. This one connects your assistant to a whole marketplace: 195+ REST APIs behind a single install and a single key, from Hacker News, Wikipedia and the World Bank to GitHub, Stripe and Anthropic.
Adding an API to the marketplace makes it callable here immediately, with no client update.
Many of the listed APIs are keyless and work the moment you connect. For the ones that need a provider key (bring your own key), you store that key once on OmniStream and the proxy injects it server-side on each call. The assistant only ever holds your Omni key, so an upstream credential is never pasted into a prompt or exposed in a tool result.
Add to ~/.claude/claude_desktop_config.json:
{
"mcpServers": {
"omnistream": {
"command": "npx",
"args": ["omnistream-mcp"],
"env": {
"OMNI_KEY": "omni_live_your_key_here"
}
}
}
}Get an Omni key at omnistream.skinvaults.online (free to start).
Same pattern - set OMNI_KEY as an environment variable and point the client at npx omnistream-mcp.
Variable | Default | Description |
| - | Your Omni API key (required for |
|
| Gateway base URL (override for self-hosted) |
With this MCP server connected, you can ask your AI assistant:
"What APIs are available on OmniStream?"
"Search OmniStream for CS2 skin price APIs"
"Get the endpoints for the skinapi API"
"Call skinapi/items-search with q=ak47 and show me the results"
OmniStream - the marketplace this server talks to, with a browsable list of every API, live uptime and a try-it console
omnistream-sdk - the same marketplace from TypeScript or Python
omnistream-go - the Go client
MIT
Call any published API on the OmniStream marketplace using the proxy. Your Omni key (OMNI_KEY env var) authenticates the request - you never need the upstream API key. Use get_api first to learn the available endpoints and their parameters.
Inputs: slug (required), operation (required), body, params.
Get full details for a specific API: description, base URL, authentication method, all endpoints with their HTTP method, path, parameters, and summary. Use this before calling call_api to understand what parameters an endpoint needs.
Inputs: slug (required).
Get details for a specific real-time stream including description, update interval, and the last 5 data samples to understand the payload shape.
Inputs: stream_id (required).
List all published APIs on the OmniStream marketplace. Returns slug, title, category, endpoint count, rating, and uptime for each API. Use this to browse what is available before calling search_apis or get_api.
Inputs: category.
List all endpoints for a specific API with their operation IDs, HTTP methods, paths, and required parameters. Lighter than get_api when you only need the endpoint list.
Inputs: slug (required).
List all available real-time data streams on the OmniStream Grid. Returns stream ID, title, category, update interval, and current subscriber count.
Inputs: no named parameters.
Search the OmniStream marketplace by keyword using BM25 ranking. Searches across API titles, descriptions, tags, and endpoint names. Returns ranked results with relevance scores.
Inputs: query (required), limit.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 7 tools
Each tool has a distinct purpose: call, get details, list, search, and stream-specific operations. No two tools overlap in functionality.
All tools follow a consistent verb_noun pattern using snake_case (e.g., list_apis, call_api), making them predictable and easy to understand.
With 7 tools, the set is well-scoped for an API and stream marketplace, covering browsing, inspection, and invocation without excess.
The tool set covers key operations for APIs (list, search, get details, list endpoints, call) and streams (list, get details). Minor gaps like lacking filtering or pagination options prevent a perfect score.
Discover and call 10,000+ production APIs from one MCP server. Pay-per-call billing for AI agents.
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
Hosted MCP server for live public-data APIs and Skills for AI agents.
