Basecamp
DownloadGet Started
ProductSandboxesEnvironmentsStacksDocsDownloadGet Started
Basecamp 0.5.0Environment bootstrap layer↗

Develop anything, faster

Ready sandboxes.
Setup disappears.

Basecamp prepares complete, reproducible development environments before execution — across languages, tools, assets, local machines, CI, and sandboxes.

Get Started →See the model ↗
Works with your stack Cloud or self-hosted Built for humans and AI
basecampsetup phase

$ basecamp sync

  • ✓ Detecting project stack...
  • ✓ Restoring runtimes & SDKs...
  • ✓ Materializing packages...
  • ✓ Fetching browsers & tools...
  • ✓ Preparing model & data assets...
  • ✓ Binding environment...
✓ Sandbox ready

$

⌘
RepositoryYour code · any language
↓
▤
basecamp.lockReproducible environment
↓
◇

RuntimesLanguages & SDKs

⬡

PackagesNative dependencies

⌕

ToolsCLI · browsers · MCP

◫

AssetsModels · data · indexes

↓
▶
Ready SandboxStart useful work immediately

Polyglot by design

One environment lifecycle.
Every stack.

Basecamp’s model is bigger than one package manager: runtimes, packages, tools, SDKs, browsers, AI assets, and data all fit into the environment graph.

Explore stack support →
Rs
Rust / CargoSystems · WASM · CLI
Shipping
JS
Node.js / pnpmWeb · full-stack · tools
Adapter roadmap
Py
Python / uvData · AI · scripts
Adapter roadmap
Go
GoCloud · CLI · services
Adapter roadmap
JVM
JVMJava · Kotlin · Scala
Model target
▰
Docker / OCIContainers · images
Model target
◉
Browser / MCP / CLIPlaywright · MCP servers
Model target
GPU
CUDA / Native SDKsGPU · ML · native libs
Model target
AI
Model AssetsWeights · tokenizers
Model target
DB
Data / IndexesDatasets · vector indexes
Model target
Shipping in v0.5Adapters and broader resource families extend the same lock → sync → activate lifecycle.

From chaos to clarity

Eliminate environment friction.

Stop spending the beginning of every run on setup, version conflicts, repeated downloads, and missing system dependencies.

BeforeEvery machine rediscovers setup.
×Install system packages$ sudo apt install ...
×Install Python dependencies$ uv sync / pip install ...
×Install Node dependencies$ pnpm install
×Install browser binaries$ npx playwright install
×Install CUDA / GPU libraries$ conda install cuda ...
×Download model assets$ huggingface-cli download ...
×Configure registries & tools$ docker login ...
×Set environment variables$ export ...
×Repair version conflictsretry / patch / repeat

▲ Environment drift · broken builds · lost time

AfterOne environment contract.Basecamp

$ basecamp sync

  • ✓ Detect project inputs and environment identity
  • ✓ Restore runtimes and toolchains
  • ✓ Materialize native package layers
  • ✓ Fetch required tools and verified assets
  • ✓ Bind the environment to this sandbox
  • ✓ Sandbox ready
↯

One commandEnvironment sync

◷

ReusableShared immutable layers

✓

ReproducibleSame lock everywhere

Get started in three steps

From repository to a working environment.

The lifecycle is the same whether the destination is a laptop, CI worker, managed agent sandbox, or isolated runtime.

01

Initialize

Connect the project once. Basecamp reads native files and creates the wider environment lock.

$ basecamp init
⌘repo
→
▤basecamp.lock
02

Sync anywhere

Resolve and materialize the runtime, packages, tools, SDKs, and assets the environment needs.

$ basecamp sync
☁
  • Runtimes
  • Packages
  • Tools
  • Assets
03

Work normally

Hand over a ready environment. The engineer or coding agent uses the project’s native commands.

$ basecamp shell
/>_
  • ✓ Code
  • ✓ Run
  • ✓ Test
  • ✓ Build

Built for AI development

More than runtimes.
Everything agents need.

AI workloads often need much more than language packages. Basecamp’s environment model is designed to make those inputs explicit before the agent starts.

AI environment model →
Ready environmentresolved before execution
◎
BrowsersChromium · Firefox · Playwright
⌘
MCP ServersModel Context Protocol tools
>_
CLI ToolsGit · ffmpeg · ripgrep · more
GPU
GPU SDKsCUDA · ROCm · native stacks
AI
Model AssetsWeights · adapters · tokenizers
T
Tokenizerstiktoken · sentencepiece · more
DB
DatasetsTraining · eval · fixtures
◇
IndexesVector data · search indexes
◎
BrowsersChromium · Firefox · Playwright
⌘
MCP ServersModel Context Protocol tools
>_
CLI ToolsGit · ffmpeg · ripgrep · more
GPU
GPU SDKsCUDA · ROCm · native stacks
AI
Model AssetsWeights · adapters · tokenizers
T
Tokenizerstiktoken · sentencepiece · more
DB
DatasetsTraining · eval · fixtures
◇
IndexesVector data · search indexes

Build without setup drag

Same environment.
Any project. Any stack.

Get Started →Read the docs ↗