Safety layer for AI-built work

Build with AI without losing the plot.

Sacred Timeline gives Codex, Claude Code, Cursor, and other AI agents checkpoints, safe experiments, rollback, backup, and a plain-English story of what changed.

Protect this project See the 90-second loop
No git knowledge required Underneath, it uses git. On the surface, you use human language.
Built for coding agents CLI, Claude/Codex skill, JSON output, MCP server, and Codex plugin manifest.
Made for real work Proposals, microsites, internal tools, strategy docs, and client assets.
The Problem

AI can change a project faster than you can understand it.

That is powerful until the working version disappears into a blur of generated edits. Sacred Timeline is for the moment when you need confidence, not another technical lecture.

You lose the trail

Files change across the project and the agent moves on before you know what happened.

Risky ideas feel dangerous

You want to try a bolder version, but not if it wrecks the one that already works.

Rollback feels technical

Git can solve this, but most AI builders do not want to become git operators.

The Safety Loop

One human vocabulary for every AI-building session.

Sacred Timeline wraps the useful parts of git in words that make sense while you work with an agent.

You want to
Say this
What happens
Save a useful moment
capture
A checkpoint is created with a plain-English message.
Try a risky idea
experiment
A safe alternate version is created so the main work stays protected.
Accept or reject the idea
keep / discard
The experiment either joins the main timeline or disappears cleanly.
Store work safely
backup
Your protected work is sent to cloud.
Understand what happened
narrate
Recent work becomes a readable story.
The 90-Second Loop

Before the agent changes anything, protect the timeline.

The demo is simple: check status, capture a working moment, start an experiment, keep or discard it, then narrate and back up the work.

1. Check

sacred doctor shows whether the project is protected and connected.

2. Capture

sacred capture saves useful progress before the agent keeps moving.

3. Experiment

sacred experiment lets you try a bold direction safely.

Sacred Timeline running in Claude Code
Who It Is For

People already letting AI agents touch real work.

Consultants and strategists

Protect proposals, client microsites, research, and operating-system work.

Founders and operators

Build internal tools with Codex or Claude without losing the working version.

Vibe coders and agencies

Add checkpoints, rollback, and a client-ready work story to fast AI-generated delivery.

Start Here

Paste one prompt into your AI agent.

You do not need to understand the terminal. Ask the agent to protect the project before changing it.

Universal starter prompt 
Before we change this project, I want to protect it with Sacred Timeline.

I do not know git, so use plain language. Treat Sacred Timeline as the safety layer for this AI-built work.

Please do this:

1. Check where we are working.
2. Check whether Sacred Timeline is installed with: command -v sacred
3. If it is not installed, install it with:
   curl -fsSL https://raw.githubusercontent.com/suhitanantula/sacred-timeline/main/install.sh | bash
4. Run: sacred doctor
5. Run: sacred status
6. If this folder is not protected yet, run: sacred start
7. Explain what you found in plain English.
8. Before making changes, tell me whether we should capture the current state or start an experiment.

From now on, use Sacred Timeline language:
- capture = save this moment
- experiment = try something risky safely
- keep = make the experiment part of the main version
- discard = abandon the experiment
- backup = send protected work to cloud
- latest = get protected work from cloud
- narrate = tell me the story of what changed

Do not teach me raw git unless I explicitly ask.