Docs home page
Search...
⌘K
Get started
Overview
Quickstart
Products
Dashboard
Logs
Traces
Prompt management
Quickstart
Prompt playground
Create prompts
Version control
Use prompt in code
Team collaboration
Evaluations
Gateway
Users
Notifications
Admin
Keywords API keys
LLM provider keys
Team management
Resources
What is LLM monitoring?
Automatic retries
How streaming works
Retrieval
Text generation
Other
Cookbooks
Build and Scale AI Products with Keywords AI: Platform Overview
How to structure LLM output with Anthropic models
Filters in Jinja templates
Discord
platform
Docs home page
Search...
⌘K
Discord
platform
platform
Search...
Navigation
Prompt management
Version control
Documentation
Integrations
API reference
Changelog ↗
Documentation
Integrations
API reference
Changelog ↗
Prompt management
Version control
Copy page
A guide to manage and track changes to your prompts.
Copy page
What is version control
Version control is a system that allows you to track changes to your prompts. It’s a way to manage and collaborate on your prompts with your team.
Why version control
Track all changes to your prompts over time
Understand who made each change and why
Roll back to earlier versions if something breaks
Work on prompts with teammates without overwriting each other’s work
Test new prompt ideas without risking your working versions
Keep a history of what worked and what didn’t
Quickstart
Version history
You can view the version history of a prompt in the Overview panel of Prompts.
If you want to see a more detailed history of a prompt, you can go to the Editor panel and click the
Version
button.
Commit changes
You can commit changes to a prompt by clicking the
Commit
button in the Editor panel.
Click on each version to see the changes you made to the prompt.
Was this page helpful?
Yes
No
Suggest edits
Raise issue
Previous
Use prompt in gateway
Deploy a prompt to your codebase and understand override modes.
Next
On this page
What is version control
Why version control
Quickstart
Version history
Commit changes
Assistant
Responses are generated using AI and may contain mistakes.