Save and version prompts

Save versioned prompts from the Playground to Prompt Hub for easy tracking and comparison.

Every save creates a new version in Prompt Hub. Browse the history, roll back if something regresses, or tag a version as Production so your application always pulls the latest approved prompt without a code deploy.

Why save and version prompts?

The Arize Prompt Hub turns prompt management from scattered snippets in Slack or code comments into a disciplined workflow: a centralized, search-friendly home for your prompt ecosystem, with immutable version control so you can track how prompts evolve over time.

• Unified prompt ecosystem - Stop hunting for the “real” prompt in tickets or repos. Store every Prompt Object - template variables, model parameters, tool definitions, and output shape - in one searchable place that keeps past, present, and future versions together.
• Immutable version control - Treat prompts with the same rigor as code. Each refinement is a hashed, immutable version you can roll back, reproduce, and tag (for example prod-v1.2) for deployment without guessing which string shipped.
• Rapid experimentation and comparison - Load any version into the Playground, test against real-world spans or golden datasets, and use side-by-side comparison to diff outputs and confirm improvements before you move a label.
• Compatible, easy-to-use SDK - Push and load prompts from your services so runtime stays in sync with Hub (see the By Code tab for links into the Python, TypeScript, and CLI Prompts API docs).
• Path to automated optimization - Once versions and eval feedback live in AX, you can move underperforming prompts into Prompt Learning feedback loops and meta-prompting to generate and rank better variants (start from Optimize a prompt).

Saving and versioning is how those benefits land in your workspace: each Save Prompt writes the next immutable version your team can diff, tag, and resolve by label at inference time.

For the full Prompt Hub story, see Why save and version prompts?.

Workflow

By Skills

Coming soon!

By Alyx

In the Prompt Playground, you can ask Alyx to persist the prompt you have been editing - for example:

Save this prompt to Prompt Hub as support-agent-v2 with a description that we tightened escalation rules.

Alyx can help you refine messages, parameters, and tools first; when you are ready, ask it to save (or click Save Prompt yourself) so a new immutable version is written to Prompt Hub with the name, description, and tags you want. You can still open Prompt Hub afterward to compare versions, add notes, or tag Production.

By UI

1. Save from the Playground

   Any time you click Save Prompt in the Playground, the current state (messages, parameters, tools, and response format) is saved as a new version in Prompt Hub. Give it a name, description, and optional tags to make it findable later.

   From the Prompt Playground, you can also use the Playground Agent to refine wording or structure before you save—then use Save Prompt to persist a new version.

2. Browse and compare versions

   Every refinement creates an immutable, hashed version. With version control, you can:

   • Compare and revert easily. If a new edit performs worse, roll back to a previous version instantly.
   • Collaborate safely. Multiple team members can test variations of the same prompt without overwriting each other’s work.
   • Maintain traceability. Each version can include notes describing what changed and why, helpful for debugging and auditing.
   • Measure improvements. Track performance across prompt iterations to identify which modifications lead to better model responses.

   View all versions of a prompt in Prompt Hub by clicking the prompt name and browsing the version history. You can tag specific versions (for example, prod-v1.2) for deployment without ambiguity.

   For a deeper walkthrough, see Save and version prompts.

3. Tag Production for deployment

   Tag the winning version as Production in Prompt Hub. The Prompt Hub SDK automatically pulls the production-tagged version at run time. This enables a human-in-the-loop workflow: you review, approve, and tag before anything goes live. Your application continuously uses the strongest prompt without code changes.

By Code

Step	Python	TypeScript	CLI
List versions	List versions	List versions	ax prompts list-versions
Create a new version	Create a new version	Create a new version	ax prompts create-version

Next up

To save and reopen whole Playground setups (not prompts), use Playground views.