SkillFoxx

Tensorlake CLI

Tensorlake offers MicroVM cloud sandboxes for agents: the tl CLI and SDK run AI code in isolation with snapshots and auto-suspend.

  • 1,044 ★ on the repository
  • Last commit
  • Apache-2.0
CLI

Extended access

Why this level

  • Agent code runs in a third-party cloud service and the access token is stored on your machine.
  • Until a network policy is set, the sandbox has unrestricted outbound internet access.
All reasons and checks

tensorlakeai/tensorlakeAuthor:tensorlakeai

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add cli/tensorlake

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Agent code runs in a third-party cloud service and the access token is stored on your machine.
  • Until a network policy is set, the sandbox has unrestricted outbound internet access.

This is third-party code. Review the repository files before installing.

What it does

Tensorlake creates MicroVM sandboxes where an agent runs tools and model-generated code. The `tl` command-line program creates a sandbox (`tl sbx create`), runs commands (`tl sbx exec <id> -- sh -lc "..."`), copies files (`tl sbx cp`), opens a terminal (`tl sbx ssh`) and ends the work (`tl sbx terminate`). The same operations exist in the Python and TypeScript SDKs. Sandboxes keep state: memory and filesystem snapshots, cloning of running sandboxes, automatic suspend when idle and resume in under a second. Outbound internet access can be closed completely (`--no-internet`) or limited to chosen addresses (`--network-allow api.example.com`). Separately there is a serverless runtime for long agent orchestrations with fan-out. You need an API key from cloud.tensorlake.ai and `tl login`.

Who it is for. Developers of agentic applications who must run model-written code in isolation and resume sessions without losing state.

Good fit when

  • An agent writes a data analysis script and running it on your own machine is risky.
  • A sandbox session must be saved as a snapshot and resumed hours later.
  • Dozens of identical copies start from one base sandbox for parallel tasks.
  • An agent's outbound traffic is limited to an allowlist of addresses.

Not a fit when

  • You work offline only: sandboxes live in the Tensorlake cloud and need an API key.
  • You need an arbitrary Docker image: only registered sandbox images are supported.
  • A couple of scripts run fast on your own machine and cloud isolation is overkill.

Example request

Create a sandbox with `tl sbx create`, run my analysis script in it, collect the result and close it, allowing outbound access only to api.example.com.

Limitations

Sandboxes run in the Tensorlake cloud and need an account and an API key, with the paid part described on the project site. The `tl` command-line program ships as a separate binary through an install script, not through PyPI or npm. A sandbox image is set by the name of a registered image, arbitrary Docker image references are not supported. Live CPU and memory changes are available only for CPU CAS sandboxes. Tokens are stored in the system keychain, and without one in the `credentials.toml` file. The last commit dates from November 2026.

How to disable. End sandboxes with `tl sbx terminate <id>`, then `tl logout --all`. Remove the SDK with `pip uninstall tensorlake` and delete the `tl` binary by hand.

FAQ

How do I install the SDK and CLI?

The SDK with `pip install tensorlake`. The CLI `tl` separately: `curl -fsSL https://tensorlake.ai/install | sh`, then `tl login` with a key from cloud.tensorlake.ai.

How do I run a command in a sandbox?

Create it with `tl sbx create`, then `tl sbx exec <id> -- sh -lc "printf hi"`. The SDK has `sandbox.run("sh", ["-lc", "..."])`.

How do I close internet access?

`tl sbx update <id> --no-internet` blocks outbound traffic, `--network-allow api.example.com` leaves chosen addresses and `--clear-network` removes limits.

What about snapshots?

A snapshot stores the memory and filesystem of a sandbox, and cloning starts a copy of a running sandbox on another machine.

Can I work with several projects?

Yes. `tl login --context staging` saves a separate context and `tl context use staging` switches it without a new login.

Discussion

No questions or reviews yet.

CLI tools

Tako VM runs agent code in single-use Docker containers with a job queue, run history and network isolation

CLIBroad access90

CLI that runs commands and AI agents in isolated microVMs that boot in under 125 ms and keeps API keys out of the sandbox itself.

CLIBroad access62
Official

A platform on your own server that gives each agent its own Linux box with SSH and MCP tools for commands and files

MCP serverBroad access318
Official

Open Terminal gives AI agents a computer of their own with a REST API: shell, files, code, browser and desktop

CLIBroad access3.6K

CLI for microVMs: an isolated virtual machine with its own kernel for agents, sub-second boot, checkpoints and branching.

CLIExtended access7.1K
Editors’ pick

Isolated Linux microVMs for agent code: the official Vercel skill on the SDK, network rules and snapshots.

SkillExtended access224

README in short

An infrastructure platform for agentic applications with Python and TypeScript SDKs, the `tl` CLI and a serverless runtime. Docs live at docs.tensorlake.ai.

README badge

Report a problem