Versioning Documentation - PromptAsCode

Overview

The Versioning tool brings Git-like version control to your prompts. Track changes over time, create branches for experiments, tag stable versions, and restore any previous state. Each version tracks not just the text, but also token counts and metadata.

Full History

Every change tracked

Branching

Experiment safely

Tags & Releases

Mark stable versions

How to Use

1
Create a Prompt - Enter your prompt text and give it a name. This creates your first version automatically.
2
Make Changes - Edit your prompt. Each save creates a new version in the history.
3
Write Commit Messages - Describe what changed and why. Good messages make history useful.
4
View History - See all versions with their timestamps, token counts, and commit messages.
5
Restore Versions - Click any version to restore it. The current version becomes a new entry in history.

Key Concepts

Versions (Commits)

Each saved state of your prompt. Includes the full text, timestamp, author, commit message, and metadata like token counts. Versions are immutable - you can't edit history.

Branches

Parallel lines of development. Create a branch to experiment without affecting the main prompt. Merge successful experiments back, or discard failed ones.

Restoration

Go back to any previous version. Restoration doesn't delete history - it creates a new version that matches the old one, preserving the full timeline.

Branching

Branch Workflow

main (production prompt)
 │
 ├── experiment-shorter
 │    └── Testing 30% token reduction
 │
 ├── experiment-tone
 │    └── More formal language
 │
 └── feature-json-output
      └── Adding structured output

When to Branch

Create a branch when:

Testing significant changes to a production prompt
Trying multiple approaches to solve a problem
Working on a feature that isn't ready for production
Collaborating with team members on different improvements

Merging Branches

When an experiment succeeds:

Compare the branch to main using Prompt Diff
Review all changes and test thoroughly
Merge the branch into main
Tag the merge as a new release if appropriate

Tags & Releases

Tagging Convention

Recommended tag naming:

v1.0.0 - Major releases with significant changes
v1.1.0 - Minor releases with new features
v1.1.1 - Patch releases with bug fixes
production - Currently deployed version
stable - Last known good version

Release Checklist

Before tagging a release:

Run the Linter (score 85+)
Test in the Playground
Review with Prompt Diff
Run benchmarks for consistency
Document changes in commit message

AI Expert Use Cases

Production Prompt Management

Maintain production prompts with full history. When issues arise, quickly identify which version introduced the problem and restore a working version.

A/B Testing

Create branches for each test variant. Track results alongside versions, then merge the winning variant back to main with documented performance data.

Compliance & Auditing

Maintain a complete audit trail of all prompt changes. Essential for regulated industries where you need to explain AI behavior at any point in time.

Team Collaboration

Multiple team members can work on different branches simultaneously. Merge changes together and resolve conflicts with the diff viewer.

Tips & Best Practices

Pro Tips

Write descriptive commit messages explaining WHY changes were made
Tag every production deployment for easy rollback
Use branches for experiments, keep main stable
Review diffs before merging branches
Include performance metrics in commit messages

Commit Message Format

Short summary (50 chars or less)

Detailed explanation if needed:
- What changed
- Why it changed
- Performance impact (if measured)

Metrics: +5% accuracy, -10% tokens