DrejT/alineo
DrejT/alineo has 43 GitHub stars and 2 forks. GitHot measured 0 stars in 24 hours and +2 in 7 days. Last checked 2026-11-10 03:42 UTC. Last push 2026-10-31.
Control plane for swarms of sandboxed AI agents: run, fork, pause, steer, and checkpoint agents that each work inside their own isolated sandbox. TypeScript SDK, daemon, and CLI.
Measured star growth
- in 24h
- 0
- in 7 days
- +2
- in 15 days
- Update delayed
- in 1 month
- Update delayed
Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.
README.md
Original
A swarm code review driven through alineod: one checkout, three forked reviewers, one merged report. Sped-up replay of a real run of the swarm-code-review cookbook.
Quickstart
bunx alineo-cli init # starts OpenSandbox in Docker
bun add alineo @alineo-labs/sqlite
export NVIDIA_API_KEY=nvapi-... # free key from build.nvidia.com
import { Alineo, textOnly, type AgentSpec } from "alineo";
import { SQLiteAdapter } from "@alineo-labs/sqlite";
const spec: AgentSpec = {
name: "hello-agent",
harness: "pi",
harnessVersion: "1.2.0",
provider: "nvidia",
model: "nvidia/nemotron-3-super-120b-a12b",
packages: ["python3"],
env: { NVIDIA_API_KEY: "${NVIDIA_API_KEY}" }, // resolved from your shell
resources: { cpu: "1000m", memory: "2Gi" },
};
const adapter = new SQLiteAdapter("./.alineo/ledger.db");
const agent = await Alineo.start(spec, { adapter });
try {
for await (const chunk of textOnly(agent.prompt("Write and run a Python hello world script."))) {
process.stdout.write(chunk);
}
} finally {
await agent.close();
}
Agents run Pi inside OpenSandbox containers. Full quickstart →
What you get
- Fork a live sandbox — spawn a child agent from a running one, with the checkout and installed packages already on disk.
- Steer, pause, resume — redirect or freeze an agent mid-turn, no restart.
- Survive a crash — alineod replays its ledger and reattaches to in-flight turns, so the daemon can die without losing the swarm.
- Gate what agents can do — hold risky tool calls and network egress for human approval.
- Audit everything — every exec and tool call lands in a durable SQLite or Postgres ledger.
Packages
| Package | Description |
|---|---|
alineo |
Run Pi coding agents in sandbox containers |
alineo-cli |
Local setup, spec management, agent session lifecycle |
@alineo-labs/sandbox |
Sandbox client — Sandbox, SandboxHandle, ExecHandle |
@alineo-labs/workflow |
Lazy pipeline builder — retry, branching, fan-out, parallel |
@alineo-labs/sqlite |
SQLite storage adapter (local dev, zero infra) |
@alineo-labs/postgres |
Postgres storage adapter (production) |
@alineo-labs/otel |
OpenTelemetry hooks adapter |
@alineo-labs/flue |
Flue runtime adapter — run Flue workflows against a sandbox |
Agent Skills
alineo ships a SKILL.md at .agents/skills/alineo/ — a curated SDK/CLI reference that AI coding
agents can load directly, installable with the Skills CLI:
npx skills add DrejT/alineo --skill alineo
Agent products give you one agent. Sandbox providers give you a box. alineo is the layer between them: it runs many agents, keeps a record of what each one did, and lets a person stay in control. It is fully open source under Apache 2.0, runs locally with no account, and has no closed control plane.
Community
Discord · Issues · Contributing · Apache 2.0
Explore related projects
codecrafters-io/build-your-own-x
Master programming by recreating your favorite technologies from scratch.
561,654 starspublic-apis/public-apis
A collective list of free APIs
498,045 starsmorluto/rea
Reverse engineer anything with agents, from app behavior down to native binaries.
450,694 starsopenclaw/openclaw
The AI that really does things. Any OS. Any Platform. The lobster way. 🦞
394,314 stars