Connect your tool server
Connect to an MCP endpoint instead of API - approved tools only, no code required.
Connect MCP tool servers to AI Agent workflows
WEXTL® supports Model Context Protocol (MCP) servers [1] as tool sources for AI Agent [2] nodes. Bring your own MCP endpoint - vendor-hosted or self-run - and expose only tools you approve for the agent to call. Sits alongside 1,300+ regular app integrations [3] as another way to give an agent capabilities.
Every MCP tool call is visible in the run history [4] - the tool invoked, its parameters, and its return value - just like any other native workflow node, no separate logs to investigate.
- GDPR
- Data Encryption
- 2FA
- Local Data Region
Video slide shows an AI Agent calling MCP tools on WEXTL® canvas.
How MCP servers support can transfrom automation?
Designed for teams that need extra capabilities for agents - to reach internal or external systems using Model Context Protocol. Here are 4 design points customers love [5] about WEXTL MCP® client support.
Auto called by model
MCP integrates with AI Agent node like the model - not a hidden plugin panel in another product.
Extends existing integrations
Internal MCP tools and regular app nodes share one workflow run and one audit trail.
Only selected tools
Select tools that AI will see, then branch on high-impact calls for human review before they fire.
How to connect MCP server to AI Agent node?
Only 3 main steps to follow - and AI Agent [2] will have access to MCP tools. Under 10 minutes [5] for the whole setup - from pointing at a MCP endpoint to a live agent calling approved tools inside your MCP automation workflow.
STEP 1
Connect using MCP server endpoint
Select from available list of supported MCP servers or use the server URL from your vendor. WEXTL retrieves tools list from it and shows exactly the tools it advertises [1] - nothing is exposed to the AI Agent [2] without your approval.
STEP 2
Select the tools for AI Agent
Pick the tools you want the agent to expose. Only approved tools are visible to the agent node [4] - the rest are invisible.
MCP tools
Select tools that AI Agent will be able to call.
list_issues
add_comment
create_issue
delete_repo
STEP 3
Route workflow based on output
Add a Condition or Merge node after the agent to gate high-impact tool calls with Human in the Loop [6] review. Each MCP tool call is logged on run history [4] with parameters and return values.
How does MCP integrated with regular apps in automation?
MCP tools have the same audit trail as regular app steps [4]. When the AI Agent [2] calls an MCP tool, the run and its result are stored inside the workflow history - same as it is logged for regular nodes.
On the canvas [7], automator sees the MCP node always connected to the AI agent node. Same drag and drop experience as with any other native app integration. And there is no limit on how many MCP nodes can be connected to AI Agent.
WEXTL® apps catalog [3] lists native app connectors that WEXTL builds and maintains, including the auth flow. For the MCP server - customers can bring their own [1] - vendor-hosted, or select from the list of available MCP servers that WEXTL supports natively.
Video shows MCP node connected to AI Agent alongside app nodes on WEXTL®.
How to guard high-impact MCP tool calls in WEXTL?
From list of available tools of your MCP endpoint select only ones that you want to let AI to call [5]. The general recommendation here - use minimal permissions that AI agent needs to perform the task with success.
Connect MCP agents to Condition, Merge and Human in the Loop nodes [6] - to have controls over outputs from the agent. For desctructive actions - you can approve tool calls manually.
Every tool call from MCP is logged inside run history [7]. Hidden bypasses are not possible and not part of the design [5].
MCP tool picker with allow-list and deny-list options on WEXTL®.
1300+ apps you can mix with MCP tools
Any app in WEXTL® Apps Library [3] can act in the same workflow run as your MCP tool calls - one canvas, one audit trail, no context handed off to a separate product.
Supported integrations
Frequently asked questions
Model Context Protocol servers advertise their tools through a tools/list call. WEXTL's AI Agent node acts as an MCP client: connect a server, and WEXTL shows exactly the tools that call returned — you pick which ones the agent may use.
A JS Code step is a fixed point in the workflow that runs the same transform every time. An MCP tool call is chosen by the AI Agent at inference time, per conversational turn — which tool fires (or whether one fires at all) can change run to run based on what's being asked.
Yes. The AI Agent node loops until done — it can call one MCP tool, feed its result back to the model, then call another (from the same or a different MCP server) based on what the response returned. Each call shows up as its own step in run history so you can follow the chain the model chose.
The step fails with a logged error, same as any node. Branch on that failure to alert the server owner, retry later, or fall back to a human path instead of the run silently stalling.
Tools the server advertises are refreshed on connect. If the server removes a tool your agent was approved to call, the agent simply stops seeing it and can no longer invoke it — no silent fallback. If the server adds a new tool, it appears in the picker so you can decide whether to expose it, but it stays unexposed until you approve it explicitly.
Connect the tools your agents need or explore more
References
List of resources used for writing this documentation page on WEXTL®. Follow a citation to jump straight to the source.
- WEXTL® MCP node — Model Context Protocol tool-server integration node.
- WEXTL® AI Agent — node documentation.
- WEXTL® Apps Library — full catalog of supported app integrations.
- WEXTL® built-in workflow nodes — catalog of built-in nodes with per-node documentation pages.
- WEXTL® anonymized usage experience from customers — collected by support team and/or partners.
- WEXTL® Human in the Loop node — pause a run for human review or approval before proceeding.
- WEXTL® visual workflow builder — canvas for building and visually tracing automations.