Building AI agents with FlowRunner: designing agent workflows, connecting models and tools, human-in-the-loop approvals, and sharing what works.
Use this category for anything related to building and running AI agents on FlowRunner, including:
- Designing agent workflows and deciding what belongs in a flow vs. an agent step
- Connecting LLMs and exposing FlowRunner actions or flows as agent tools (including MCP)
- Prompting, tool selection, and getting reliable agent behavior
- Human-in-the-loop approvals and guardrails for agent actions
- Debugging agent runs: unexpected tool calls, loops, errors, and costs
When asking for help, describe what the agent is supposed to do, which model and tools it uses, and what actually happened (run output or error). Remove API keys and any sensitive data from prompts or logs before posting. If the problem is with a specific integration’s action rather than the agent logic, post in Extensions instead.