Plug MCP Server icon

Plug MCP Server

Expose governed Plug capabilities to AI agents through an MCP-style tools/list and tools/call contract.

Actions3

Call Capability

AI-generated

Summary

Execute a defined capability in the Plug MCP Server by specifying its name and parameters to perform various governed operations such as queries or tool executions.

Inputs

  • Authoring Mode — Select how to author capability definitions: as raw JSON or via the Visual Builder form.
  • Capability Definitions JSON (required) — Provide the capability registry as JSON defining capabilities' semantic contracts, governance rules, and execution configurations (used when Authoring Mode is JSON).
  • Capabilities (Visual Builder) (required) — Define capabilities via structured form fields as an alternative to JSON authoring (used when Authoring Mode is Visual Builder).
  • Capability Name (required) — Select the technical name of the capability to execute from the defined registry or set via expression.
  • Params Input Mode (required) — Choose how to input capability parameters: as a JSON object or via mapped typed fields.
  • Capability Params JSON — Pass business parameters as a JSON object matching the selected capability's schema (used when Params Input Mode is JSON).
  • Capability Params (Mapped Fields) — Fill typed input fields generated from the capability's parameter schema (used when Params Input Mode is Mapped Fields).
  • Audit User ID — User identifier to record in the MCP audit log for traceability.
  • Audit Session ID — Session identifier to correlate audit logs for conversation turns; auto-generated if left empty.
  • Include Audit In Output — Enable to attach sanitized audit data to the output item; disable when forwarding responses directly to AI agents.
  • Forbidden Capability Names JSON — List capability names to exclude from invocation and listing for this agent profile.
  • Max Tool Calls Per Turn — Optional limit on the number of allowed tool calls in the current conversation turn; 0 inherits limit from AI Hub wiring if present.
  • Tool Call Count — Current ordinal number of the tool call within the turn, used for enforcing Max Tool Calls Per Turn.
  • AI Hub Node Name — Optional reference to a sibling Plug AI Hub node to obtain governance data if not provided by the current input item.
  • Agent ID — Advanced option to override the agent ID used during execution.
  • Client Token — Advanced option to override the client token used during execution.
  • MCP Protocol Version — Advanced option specifying the MCP protocol version for compatibility.

Output shape

a single JSON object representing the execution result of the called capability, optionally including sanitized audit data.

The output includes the capability execution result compliant with governance rules defined in the capability registry. Audit information is included if enabled. Errors related to forbidden capabilities or governance violations are reported in the output.

Links

Discussion