Connect agents to MCP tools
Let the agent discover and call tools exposed by MCP servers, so you can extend chat beyond Q&A to complete real tasks.
Orchestrate AI chat with MCP tools
Use this blueprint to run an AI-driven chat in Camunda where an AI agent can discover and call tools exposed by Model Context Protocol (MCP) servers. The process uses an ad-hoc sub-process with the Agentic AI connector plus MCP Client and MCP Remote Client connectors for tools such as OpenMemory, filesystem, and time. For filesystem actions, the process asks for user confirmation before executing the tool call, giving you a clear human-in-the-loop guardrail pattern. This blueprint is designed for self-managed or hybrid deployments where MCP clients are configured in the connectors runtime, and it is preconfigured to run with AWS Bedrock.
Features and Benefits
Add human approval for filesystem actions
Route filesystem tool calls through a confirmation task so a user can approve or deny execution before any file operation runs.
Persist and recall important context
Use the OpenMemory MCP tool to save key facts during a conversation and retrieve them later for consistent follow-ups.
Iterate until the user is satisfied
Capture user feedback after each agent response and loop back for follow-up prompts until the user confirms the outcome is complete.
Mix local and remote tool integrations
Combine MCP Client and MCP Remote Client connectors with in-process tools, giving you a flexible foundation for governed agentic automation.
Details
- Marketplace release date -
- Last Github commit -
-
Associated Product Group Categories:
- Quick Start
-
Version Compatibility:
-
Used resources:
522488