VerveContext

One endpoint, grounds every agent

Point your agent framework at the VerveContext MCP server and every enabled source becomes a tool the model can call to check itself — no per-source plumbing. One endpoint, one scoped key, 300+ live sources your agents can cite.

  • MCP-native
  • Scoped keys
  • Audit logs
Get grounded→Read the docs
What you connect to

One MCP server, every source

Native protocol, scoped keys, audit logs, and one key across all 300+ sources.

Model Context Protocol

VerveContext is a native MCP server, so it works with any MCP client — Claude, Cursor, LangChain, or your own. Every enabled source appears as a callable tool with a model-readable name and when-to-use hints; there's no per-source plumbing to write.

  • Native MCP
  • Any client

Scoped keys

Grant an agent only the sources it should reach. A key scoped to markets and weather can't quietly call anything else — so a single connection stays least-privilege across every agent you run.

  • Least privilege
  • Per agent

Usage & audit logs

Every grounding call is recorded — which source, which key, when. See exactly what was pulled to ground each answer, so you can trace a claim back to the moment its data was fetched.

  • Per-source usage
  • Traceable

One key across every source

300+ sources ride the same endpoint and the same key. Enable markets today and add news next week without touching the connection — the tool list grows, the setup doesn't change.

  • 300+ sources
  • One endpoint
Setup

Three lines of config to grounded answers

Add the endpoint and your key to your MCP client's config — a context.json / mcpServers block naming the server url and an x-api-key header. No SDK, no per-source wiring.

  1. 01

    Point your agent at the endpoint

    Add VerveContext to your MCP client's server config — the url is https://api.vervecontext.com/v1/mcp. In most clients this is a small JSON entry (a context.json / mcpServers block) naming the server and its url.

  2. 02

    Add your scoped key

    Attach your key as the x-api-key header on that server entry. Keys look like vc_live_… and carry exactly the sources you scoped them to — nothing more.

  3. 03

    Every enabled source is a callable tool

    Restart the client and the sources you turned on show up as tools. The agent calls them when an answer needs a real value, and each fact comes back with the source it can cite.

Per-framework guides

Pick your agent

Same endpoint, different setup mechanics. Follow the guide for the framework you run.

Claude

Claude natively speaks MCP. Add VerveContext as a server in Claude Desktop and its answers start citing live sources.

  • Claude Desktop
  • mcpServers

Cursor

Wire sources in as MCP tools inside the editor, so agentic coding workflows can check real data while you build.

  • .cursor/mcp.json
  • In-editor

LangChain

Load VerveContext's MCP tools into a chain or agent and let the pipeline ground its steps in live facts.

  • Chains & agents
  • Tools
Ground ClaudeGround CursorGround LangChain

Connecting, answered.

What teams check before pointing an agent at the endpoint.

MCP setup guides →
Which agents can connect?
Any MCP client. VerveContext is a native Model Context Protocol server, so anything that speaks MCP — Claude, Cursor, LangChain, LlamaIndex, or a custom client — can point at the endpoint and use the sources as tools. There's nothing framework-specific about the server itself.
What's the endpoint and key format?
The MCP endpoint is https://api.vervecontext.com/v1/mcp, and keys look like vc_live_… passed as an x-api-key header. That single pair connects every source you've enabled.
Do I need to write a connector per source?
No — that's the point of connecting over MCP. One server entry exposes every enabled source as a ready-made tool. Enabling a new source adds it to the agent's tool list without any change to your connection.
Is there a non-agent option?
Yes. REST access is available for backends that aren't agent-based, so the same sources can ground a traditional service. The MCP endpoint is the path most agent frameworks use.
What shows on my bill?
VerveContext runs on APIVerve, our production data engine, so invoices and card statements read APIVERVE. Same account, same key, same rails — nothing else changes.

Connect once, ground everything. One MCP endpoint, one scoped key, and every enabled source becomes a tool your agent can cite.

Get grounded→How grounding works

Enterprise?

SSO, custom SLAs, private sources, and dedicated support for regulated teams.

See pricing