Open source · local first · Apache-2.0

Restore Rehearsal

Test a real backup before an upgrade reaches production—and give people and AI agents a focused, private evidence graph of the whole project.

See the workflow · 23 seconds

Know your project. Prove it can recover.

A short visual overview of the private Project Brain, isolated backup restore testing, verification, and upgrade rehearsal.

No narration. On-screen text describes each step. Music: Sascha Ende, ende.app (CC BY 4.0).

Two connected problems

A backup is not proof. A codebase is not context.

Rehearse makes recovery evidence and project understanding inspectable instead of relying on assumptions hidden in scripts, tickets, or someone’s memory.

01 · Recovery

Restore before production

Run a declared backup restore in a disposable environment, check the application before and after a candidate upgrade, retain share-safe evidence, and clean up the stack.

Explore backup restore testing →

02 · Understanding

Keep the project’s evidence connected

Statically scan a Git repository or ordinary local folder into one private SQLite file containing architecture, dependencies, tests, deployment, recovery, findings, decisions, and structural history.

Explore Project Brain →

One evidence workflow

Understand, rehearse, inspect.

Each capability works independently. Together they give a person or agent a bounded route from project structure to affected tests and recovery evidence.

Local projectSource, configuration, tests, decisions
Private brainConnections, changes, findings, evidence
Focused actionReview, test selection, restore rehearsal

Universal inspection

A consistent evidence plan for websites, native apps, CLIs, APIs, code, documentation, and data—including screenshots when an inspection tool is available.

Explicit-trust extensions

Bounded evidence importers, offline recovery-adapter proposals, a typed Python plugin SDK, and a local stdio MCP interface.

Portable Agent Skills

Nine inspectable skills can be installed into Codex-style, Claude Code, or GitHub Copilot project directories without granting new capabilities.

See the Agent Skills →

Install the standalone application

Start locally with Python 3.11+

Project Brain and the inspection, skill, importer, adapter, and MCP features do not require Git or Docker. Executing restore drills additionally requires Linux, Docker Engine, and Docker Compose v2.

uv tool install "git+https://github.com/Someone006/Restore-Rehearsal.git@v0.3.0"
rehearse --version

Prefer pipx? Clone the source and run pipx install .. Read the complete documentation before configuring a restore recipe.

Private by design

Project evidence stays local.

Rehearse collects no telemetry and does not upload project data, reports, screenshots, or backup contents. Its offline Project Brain visual uses no remote scripts, fonts, analytics, or network requests.

API keys, credentials, tokens, private keys, and backup contents must never be uploaded. Any other outbound project information requires the user’s explicit destination and scope consent.

Commercially usable

Free and open source.

Apache-2.0 permits commercial use, modification, and redistribution subject to its terms. There is no paid tier required.

Alpha limitation: Rehearse must not be the sole production release gate, disaster-recovery control, accessibility certification, security assessment, or legal/compliance decision.

Plain answers

Frequently asked questions

Does Project Brain require a Git repository?

No. It can statically scan a Git work tree or an ordinary local folder. Git-aware ignore behavior is used when Git is present.

Does the project scan execute my code?

No. Project Brain performs a bounded static scan. Restore drills are different: they execute only the reviewed commands declared in a recipe inside a disposable Docker Compose environment.

Is Restore Rehearsal backup software?

No. It does not create or replace backups. It tests a declared backup and recovery path so that the existence of a backup is not mistaken for evidence that recovery works.

Can companies use it?

Yes. Apache-2.0 permits commercial use. The current release is alpha and should be treated as an advisory tool with human review, not a sole production or compliance control.