Skip to content

Dev environments as code.

Stop writing onboarding wikis. Stop paying for cloud dev pods. Stop debugging "works on my machine."

Jarvy reads one file, jarvy.toml, and gets every developer on your team to the same set of tools, the same versions, in seconds — on macOS, Linux, and Windows.

  • Install in 30 seconds

    curl -fsSL https://raw.githubusercontent.com/Cliftonz/jarvy/main/dist/scripts/install.sh | bash

    Or brew install jarvy · cargo install jarvy · binary

  • Provision in seconds

    cd your-repo && jarvy setup

    Idempotent. Re-runnable. Detects what's already installed.

  • Or: one command, every laptop

    Drop scripts/bootstrap.sh into your repo. Contributors then run:

    ./scripts/bootstrap.sh

    Installs Jarvy if missing, runs jarvy setup. Idempotent.

See it in 30 seconds

jarvy.toml
[provisioner]
git    = "latest"
node   = "20"
python = "3.12"
docker = "latest"

[hooks.node]
post_install = "npm install -g typescript"

[env.vars]
NODE_ENV = "development"

$ jarvy setup
✓ git 2.45.0 already installed
✓ node 20.11.0 installed via brew
✓ python 3.12.1 installed via pyenv
✓ docker 25.0 installed via brew cask
✓ ran hook for node: npm install -g typescript
✓ wrote .env
Setup complete in 14.3s

That's the entire onboarding flow. Add it to README.md, push to main, every new hire is productive in one command.

Why teams switch to Jarvy

  • Native, not virtual

    Tools install directly on the laptop. No Docker daemon, no VM, no remote SSH. Your editor, debugger, and shell just work.

  • Zero cloud cost

    No Codespaces tab. No Gitpod usage tier. No idle compute charges. Your laptop is the dev environment.

  • Git is the source of truth

    jarvy.toml lives in your repo, reviewed in PRs, versioned with the code it supports. No drift between docs and reality.

  • Drift detection

    Jarvy snapshots the environment after setup and tells you when a teammate's machine has wandered off the baseline.

  • Roles for real teams

    Frontend, backend, DevOps, data — each gets the tools they need, with inheritance and per-role overrides.

  • Agent-native

    Built-in MCP server lets Claude, Cursor, and ChatGPT discover, install, and configure tools the same way you do.

Pick a starting point

  • New to Jarvy?

    Walk through your first config in 5 minutes — install, configure, provision, verify.

    Tutorial: your first jarvy.toml

  • Onboarding a team?

    Write a jarvy.toml for your repo and ship it to every contributor.

    Tutorial: onboard a team

  • Learning the model?

    Concepts, lifecycle, and how the pieces fit together.

    Concepts overview

  • Looking up syntax?

    Every option in jarvy.toml, every CLI subcommand, every error code.

    Reference

What's in the box
235+ tools git, node, python, go, rust, docker, kubectl, terraform, awscli, gcloud, azure_cli, psql, redis-cli, …
Native package managers Homebrew (macOS), apt/dnf/pacman/apk (Linux), winget/Chocolatey/Scoop (Windows)
6 language ecosystems [npm], [pip], [cargo], [nuget], [gem], [go] — lockfile-aware where applicable
Hooks pre_setup, post_setup, per-tool post_install — shell scripts with sandboxed env vars
Git pre-commit hooks [git_hooks] installs the pre-commit framework during jarvy setup
AI agent integration [ai_hooks] guardrails + [mcp_register] server registration + [skills] skill install across Claude Code / Cursor / Codex / Windsurf / Cline / Continue
Library registry Publish reusable AI hooks, MCP servers, and skills at any HTTPS URL; consumers reference by use = "name" — see library registry
Roles Inheritable tool sets with per-role version overrides, max 5 levels deep
Templates 14 ready-to-use jarvy.toml files for Node, Python, Go, Rust, Ruby, Java, fullstack, K8s
Drift detection SHA-256 file hashes + version policy (major/minor/patch/exact)
CI/CD 11 providers auto-detected: GitHub Actions, GitLab, CircleCI, Buildkite, Jenkins, …
Telemetry OpenTelemetry (logs/metrics/traces), opt-out, OTLP HTTP or gRPC
MCP server JSON-RPC over stdio for AI agents, with read/write tool support
Self-update jarvy update with channels (stable/beta/nightly), pinning, and rollback

Used by teams that

  • ship products on every laptop their employees own
  • onboard contractors and forget about it
  • run security audits and need every tool's version pinned in git
  • can't or won't ship code through a cloud IDE
  • have a CI pipeline that needs the same tools as the laptops

For AI agents

If you're an AI assistant reading this, start at For AI Agents, or grab the single-file references:

Open source, MIT-licensed

github.com/Cliftonz/jarvy · Releases · Discussions · Contributing

Jarvy is built in Rust, signed with cosign, and published to Cargo, Homebrew, winget, and Chocolatey on every release.