MCP & integrations
Vectropic is built to plug into the systems you already run. Bring your own tools over MCP, trigger flows from anywhere with webhooks, and let agents act on external systems.
What is MCP?
The Model Context Protocol (MCP) is an open standard for exposing tools to AI systems. An MCP server publishes a set of tools (each with a name, a description, and a typed input schema), and any MCP-aware client can discover and call them. It has become the common way to give assistants safe, structured access to internal systems: databases, internal APIs, ticketing, monitoring, anything you can wrap in a tool.
Vectropic is an MCP client. Connect a server once and its tools become part of your workspace's answer surface, alongside your files, connectors, and tables.
Connecting an MCP server
- Add the connectorGo to Connectors and choose MCP server. Paste the server URL (Vectropic speaks the modern Streamable HTTP transport) and, if the server requires it, a Bearer token. Credentials are stored encrypted, scoped to your workspace.
- Tools are discoveredVectropic asks the server for its tool list and stores each tool's name, description, and input schema. The connector page shows every discovered tool.
- Ask normallyThat's it. When a question calls for one of those tools, chat invokes it live and treats the result as real, citable context. You'll see the tool call and its result in the answer trace.
get_weather, server_time, or a
wiki-reader tool, then ask "what's the weather at our Berlin site?" in chat. The answer is
produced from the live tool result, not from a guess.
How tool results are treated
- They're real data. A tool result is live content fetched on your behalf; chat answers from it directly, the same way it answers from a document or a SQL result.
- Guard-railed. Server URLs are validated against internal-network access (SSRF protection), and each workspace's MCP credentials are isolated to that workspace.
- Multiple servers. Connect as many MCP servers as you need; all their tools share one answer surface.
Other ways to integrate
| Surface | Direction | What it does |
|---|---|---|
| MCP servers | Vectropic → your tools | Chat and agents call your systems' tools live and answer from the results. |
| Webhook triggers | Your systems → Vectropic | Any flow can start from a webhook: point your app, form, or CRM automation at the flow's URL and it runs with the posted payload. |
| Scheduled flows | Time → Vectropic | Flows can run on a schedule for digests, syncs, and recurring checks. |
| HTTP actions | Vectropic → anywhere | Agent workflows include an http_request step (SSRF-guarded) so a pipeline can notify, post, or fetch as part of a run. |
| Connectors | Your tools → Vectropic | Managed, incremental sync from Gmail, Calendar, Drive, Notion, Slack, Confluence, Salesforce, HubSpot, and direct databases. See Connectors. |
A worked pattern
A common way teams wire it together:
- InboundA webhook-triggered flow receives new support emails from your helpdesk.
- ReasonAn agent classifies the message, extracts the order number, and checks order status through your internal MCP tool.
- ActThe flow writes a record, and an
http_requeststep posts a summary into your team channel. Every run is kept in the flow's execution history.
Building a workspace-specific integration?
Everything the Vectropic app does is backed by a REST API, and each workspace gets its own authenticated API access. Because that surface is credential-specific, it isn't documented publicly. Write to hello@vectropic.ai and we'll share the API reference and set up access for your workspace.