Your tools, already wired in
Bring any MCP server or native integration. Agents call Slack, Linear, Stripe, Notion, and your databases — with tools scoped per workflow, not per account.
An agent is only as useful as what it can touch. Arcflux speaks MCP — the open Model Context Protocol — so the growing ecosystem of MCP servers plugs straight into your flows, alongside native integrations for the tools teams already run on.
Vendor-hosted, self-hosted, or native
- Vendor-hosted MCPs — connect with OAuth — no infrastructure on your side.
- Bring your own — any MCP server you run can be a tool source.
- Native integrations — first-class connectors where MCP isn't the right fit.
Scoped, not sprawling
- Tools are scoped per workflow — a flow only sees the tools you gave it.
- OAuth connections are managed once, reused across flows.
- Every tool call lands in the run's step-by-step audit trail.
Models included
LLM steps run on Claude, GPT, Gemini, Grok, or Groq — picked per step. Providers and tools live in the same graph, so “ask Claude, file in Linear, notify Slack” is three nodes.
webhook POST /alerts 202 → agent + runbook lookup → linear.createIssue (full context attached) → slack.notify #oncall
Common questions
The short version of how mcp & integrations works. Anything else: info@arcflux.ai.
The Model Context Protocol — an open standard for exposing tools to AI models. Any MCP-compatible server can become a tool source in Arcflux.
Not for vendor-hosted MCPs — connect with OAuth. If you run your own MCP server, point Arcflux at it.
Yes — tools are scoped per workflow. A support flow doesn't see your billing tools.
Claude, GPT, Gemini, Grok, and Groq — switchable per step.
Turn your next event into a workflow
Model agentic workflows visually, trigger them from anywhere, and govern every run.