DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to refine an existing workflow with a follow-up change with FlowX

Apply a targeted change to one node or the whole workflow without rebuilding everything.

Who it is forA developer who has a generated workflow and needs to add a step, such as input validation or a summary, without starting over.

The problemRe-prompting from scratch regenerates work that was already correct and makes changes harder to control.

The outcomeAn updated workflow in which only the affected nodes change, rerun as a service through the SDK.

What you need first

  • Python 3.10 or later
  • An LLM provider, model, and API key for FLOWX_LLM_PROVIDER, FLOWX_LLM_MODEL, and FLOWX_LLM_API_KEY
  • A local MCP-capable agent client, or the Python SDK in your own script
  • A writable folder for FLOWX_DEFAULT_WORKSPACE

Steps

  1. Run python -m pip install -e . in the FlowX folder after completing the .env setup.
  2. Create a Python script that opens FlowXClient(workspace="./workflows") and calls build_workflow_json with your process description.
  3. Call list_nodes on the workflow ID to get each node's step_id and node_id.
  4. Call evolve(workflow_id, "Validate empty input.", target=nodes[0]["step_id"]) to change one node, or evolve without a target to let the engine decide.
  5. Call run_service(workflow_id) to start or restart the workflow as a service.

Prompt for your agent

Using the FlowX Python SDK in the repo at <repo-path>, open the workflow <workflow-id>, list its nodes, and apply the change: Validate empty input. to the node <step-id>. Report the list of nodes before and after, and confirm that run_service starts the workflow.

Caveat. The SDK examples in the README are the reference for the calls shown here, and the walkthrough scripts are linked from the README rather than reproduced in full.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for AIpRoBuilder/FlowX and "Refine an existing workflow with a follow-up change"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer