Jev State

About Jev State

Jev State is a free, open-source developer tool that converts AI conversations into tests and runnable code. It lets you build conversational workflows, inspect why each step occurred, and save interactions as tests to catch logic errors before deployment. The tool exports runnable TypeScript, workflow JSON, and an integration skill file for coding agents.

Review

Jev State approaches conversational AI development from a testing and observability angle. Rather than focusing solely on prompt engineering, it surfaces the decision path an AI workflow takes-including wrong turns that still lead to correct answers. The ability to save a conversation as a repeatable test and export the underlying code gives developers a concrete artifact to integrate into their existing toolchain.

Key Features

• Step-level trace visibility. Jev State shows what happened at each state in a conversation, so you can see where the workflow zigged when it should have zagged-even if the final output looks correct.
• Conversation-to-test capture. Any live conversation can be saved as a test. These tests check the intermediate steps, not just the final result, which helps catch subtle routing or logic issues.
• Exportable artifacts. The tool generates runnable TypeScript, workflow JSON, and a SKILL.md file that a coding agent can use to integrate the workflow into an application.
• Dual execution modes. Live runs use your own TypeSafe key and incur provider billing. A key-free simulation mode is also available for verifying wiring without making live API calls.
• JSON re-import. Workflows exported as JSON can be imported back into Jev State later for further editing.

Pricing and Value

Jev State itself is free and open source. Live execution requires a TypeSafe key, and costs are billed directly by the underlying provider. Simulation mode carries no cost. The pricing model for future managed services or hosted options has not been defined.

Pros

• Exposes the full decision path of an AI workflow, which makes debugging conversational logic more transparent.
• Tests validate intermediate steps, catching problems that final-output checks miss.
• Exports standard artifacts-TypeScript, JSON, and a skill file-that fit into a typical development workflow.
• Key-free simulation mode lets you prototype and verify wiring before incurring any API costs.
• Open-source codebase allows inspection and modification without vendor lock-in.

Cons

• The tool requires a TypeSafe key for live runs; users outside the TypeSafe ecosystem will need to set that up first.
• It is not well suited for teams that do not need step-level debugging and prefer a higher-level visual flow builder.
• As a newly launched project, community resources, third-party integrations, and long-term maintenance patterns are still unproven.

Jev State fits developers who need to test conversational AI workflows beyond surface-level outputs and want exportable code they can hand off to a coding agent. Teams building support routing, onboarding flows, or any multi-step conversational logic will find the step-level trace and test capture directly useful. It's less relevant for those who only need simple, single-turn prompt testing or who operate entirely outside the TypeScript/TypeSafe ecosystem.

Open 'Jev State' Website