I've been working hard on the recent update to my Claude Code plugin to allow it to understand existing Architect flows:
LinkedIn post about the release
This has been an interesting challenge as flows are complex and their JSON definition (which my plugin retrieves) can be jolly big! So can easily pollute an AI agent's context.
I believe I have cracked it now though, as testing by me and others in the community of v2 of the Claude Code plugin attests to.
It works by deterministically simplifying flows into a intermediate representation, which Claude Code can progressively enrich and search.
This now means you can use AI to, among other things:
- Identify logical issues - such as misinforming the customer of where they're being transferred, never leaving a loop due to incorrectly updating a variable
- Static issues - dead branches, not handling Data Action failures/timeouts
- Enforcing standards - enforcing a set of standards are followed in how the flow is created, such as consistent variable names and error handling
- Documenting flows - as well as being able to ask questions of them
- Mapping dependencies
If you have any feedback I'd love to hear it! My aim is to make developing and maintaining flows easier, so hopefully this goes some way to achieving that.
You can find the plugin here: GitHub - MakingChatbots/genesys-cloud-plugins
#Architect#CXasCode#PlatformAPI------------------------------
Lucas Woodward
Winner of Orchestrator of the Year, Developer (2025)
LinkedIn -
https://www.linkedin.com/in/lucas-woodward-the-devNewsletter -
https://makingchatbots.com------------------------------