LaunchRally

Know what stands between your repository and a trustworthy launch.

LaunchRally is a local-first, open-source launch-readiness audit and verification tool for repository-owning AI builders. It turns repository facts, declared launch intent, and explicitly approved read-only evidence into a deterministic assessment and an ordered path to verification.

It needs no LaunchRally account, uses no default telemetry, and makes no repository, deployment, or Provider write during an Audit.

Status: Stable. P0 is Product Complete and 0.3.2 is publicly available on the stable latest channel. LaunchRally is P0 Validated and the Quality Floor is satisfied.

First Audit

Install the exact Stable Launcher through your current npm prefix and verify its structured version output before entering a repository:

npm install --global @launchrally/cli@0.3.2
rally --version --json

If npm reports a prefix permission error or rally is not on PATH, do not use sudo. Follow the Install guide to choose a Node version manager or user-writable npm prefix and expose its executable directory yourself.

From the repository root, run the Human Audit and save its complete Report, then separately preview and confirm Init:

rally audit --plain --cwd . --output ./launchrally-audit-report.json
rally init --plain --cwd . --report ./launchrally-audit-report.json
rally --version --json --cwd .

Review the complete Brief before confirming the Audit. Public verification and every Provider read are independent permissions and default-denied. Denial remains a visible Verification Gap. Audit does not create .launchrally; the saved Report is an explicit output outside that directory.

Confirmed Init is the first project mutation. Its preview is limited to LaunchRally-owned .launchrally paths, materializes the exact Project Toolchain, and leaves application dependency files unchanged. The final read-only version command proves that the Launcher selected the project-pinned Engine through validated Execution Authority. A same, newer, or supported older Launcher follows that project pin instead of silently changing it.

The terminal below uses deterministic synthetic fixture output. Your result reflects your repository, confirmed scope, evidence, and permission choices.

Continue with the Quickstart for the complete journey.

No-install trial and CI fallback

Exact-version npm-exec is supported as a no-install trial and CI fallback, not the default interactive journey. Its follow-ups retain the complete prefix so they remain executable after the first process exits:

npm exec --package=@launchrally/cli@0.3.2 -- rally audit --plain --cwd . --output ./launchrally-audit-report.json
npm exec --package=@launchrally/cli@0.3.2 -- rally init --plain --cwd . --report ./launchrally-audit-report.json
npm exec --package=@launchrally/cli@0.3.2 -- rally --version --json --cwd .

How it works

1. Audit — scan local facts, confirm the complete Brief, decide each optional read, and save the complete Report. Audit does not create project state.
2. Init and delegate — preview and confirm LaunchRally-owned files, materialize the exact project Engine, and let the Launcher follow that pin for repository operations.
3. Plan and Verify — use the current Report for read-only guidance, hand remediation to your Agent only when requested, and recollect evidence for a fresh result.

The Launcher is the user-managed rally dispatcher. The Engine executes repository operations. The Project Toolchain records and materializes the exact project Engine. Execution Authority selects and validates it, while Invocation Context only describes how the Launcher was entered so safe follow-up commands can be rendered.

What it checks and produces

Area	What LaunchRally does
Repository and release intent	Discovers supported local facts, then asks you to confirm inferred scope.
Web launch baseline	Evaluates security, reliability, operability, and release-readiness Checks against versioned policy.
Public and Provider evidence	Performs only the bounded, read-only operations you approve; missing evidence stays visible.
Audit output	Returns a JSON Report Record, Markdown Report View, Evidence Index, Findings, and Verification Gaps.
Follow-up work	Produces a deterministic Launch Plan and fresh full or targeted verification results.

The Coverage Acceptance Matrix demonstrates the universal Baseline journey, not a framework or Provider support allowlist.

Safety and permissions

• No LaunchRally account, private service, default telemetry, or mandatory Report upload.
• Audit performs no repository writes; confirmed Init is preview-first and stays under .launchrally.
• Local scan, public verification, and every Provider read are distinct, default-denied authorization boundaries.
• CLI or Plugin installation grants no Provider, deployment, production, credential, or application-source write authority.
• Reports and Evidence remain local unless the repository owner explicitly moves them.
• LaunchRally never installs Provider tools, initiates login, provisions infrastructure, deploys, stages, or commits changes.

Read the complete privacy and permission model and data model before using LaunchRally with sensitive repositories.

Commands and integrations

Command or integration	Purpose	Guide
rally audit	Assess confirmed launch scope from local and approved read-only evidence.	Quickstart
rally init	Preview and confirm isolated .launchrally project adoption.	Initialization
rally toolchain	Inspect, restore, migrate, or clean the project-owned toolchain explicitly.	Install guide
rally plan	Turn current Findings into ordered, read-only remediation guidance.	Planning
rally handoff	Discover and explicitly approve one bounded external Executor authority package.	External Executor handoff
rally providers	Compare advisory Provider Decision Cards after confirming constraints.	Provider guidance
rally verify	Recollect evidence and record a fresh full or targeted result.	Verification
Codex and Claude	Install, update, remove, and validate the Agent adapters separately from the CLI.	Install guide

Packages

Package	Audience and purpose
@launchrally/cli	Builders running the first Audit and complete command journey.
@launchrally/core	Library authors embedding deterministic audit and verification logic.
@launchrally/contracts	Integrators consuming versioned schemas, constants, and protocol contracts.
@launchrally/codex-plugin	Codex users installing the canonical Agent Skill adapter.
@launchrally/claude-plugin	Claude Code users installing the canonical Agent Skill adapter.

Documentation and project links

• Documentation index
• Quickstart
• Install and release guide — lifecycle and troubleshooting authority
• Project status and Telemetry-Free Validation
• Contributing
• Security
• GitHub Discussions
• GitHub Issues
• Apache-2.0 license