Yash Shah f0045a3a5f fix(api): snapshot files only and surface HTTP errors in the SDK
Context:
- tar.add on directories recursively re-included main.py/main.js that snapshot restore is supposed to skip.
- AgentboxClient.raise_for_status turned expected 408 timeouts into opaque httpx errors.
- Missing runtimes returned HTTP 200 with exit 127 instead of 400.

Test Plan:
- docker compose run --rm test
- tests/test_api.py::test_snapshot_restore_keeps_workspace_file
- tests/test_api.py::test_run_timeout_limit

Additional Notes:
- health CLI now prints the package version, matching the other five CLIs.
- TestClient is used as a context manager to avoid the Starlette lifespan warning path.
2026-08-19 11:57:34 +05:30

agentbox

Self-hosted code execution sandbox for AI agents — one docker compose up gives you an HTTP API for running untrusted code in isolated environments.

Status: v0.3 — Python + Node subprocess sandbox, per-run timeouts, TypeScript client, workspace snapshots.

Problem

Every agent that writes and runs code needs a safe execution environment. Teams either YOLO in shared containers or pay per-second for hosted sandboxes. Self-hosting gVisor/Firecracker is weeks of work.

Key features (v0.2)

  • HTTP API: POST /v1/run executes Python or JavaScript
  • Per-request limits.timeout_seconds (HTTP 408 on timeout)
  • Workspace snapshots: "snapshot": true then "snapshot_id"
  • Python SDK + TypeScript client (sdk/ts/client.ts)
  • Docker image includes Node.js for the JS runtime
  • Credential stripping when the backend is not unrestricted

Architecture

Agent / SDK
    └── POST /v1/run
            └── SubprocessSandbox (MVP)
                    └── (next) gVisor / Docker backend
Component Technology Why
API FastAPI Async-ready, OpenAPI docs, widely adopted
Server uvicorn Standard ASGI server
Config pydantic-settings Typed env config
Tests pytest + httpx TestClient Fast API testing

Installation

pip install agentbox
pip install -e ".[dev]"

Usage

Start server

agentbox serve
# or
docker compose up agentbox

Run code

curl -X POST http://localhost:8080/v1/run \
  -H 'Content-Type: application/json' \
  -d '{"code": "print(sum(range(10)))"}'

Python SDK

from agentbox.sdk.client import AgentboxClient

client = AgentboxClient("http://localhost:8080")
print(client.health())
print(client.run("print('hello')"))
print(client.run("console.log('hello')", language="javascript", timeout_seconds=5))
snap = client.run("open('memo.txt','w').write('kept')", snapshot=True)
print(client.run("print(open('memo.txt').read())", snapshot_id=snap["snapshot_id"]))
client.close()

Docker

docker compose up agentbox        # start API on :8080
docker compose run --rm test    # run unit tests

Configuration

Variable Default Description
AGENTBOX_HOST 0.0.0.0 Bind host
AGENTBOX_PORT 8080 Bind port
AGENTBOX_DEFAULT_TIMEOUT_SECONDS 30 Execution timeout
AGENTBOX_SANDBOX_BACKEND subprocess Backend selector
AGENTBOX_SNAPSHOT_DIR /tmp/agentbox-snapshots Workspace snapshot store

Running tests

pytest tests/ -v

Roadmap

  • Node.js runtime + TypeScript client + per-run timeout
  • Filesystem snapshot/restore (tar workspaces)
  • gVisor runsc backend with warm pool
  • Default-deny egress with allowlists (kernel netns)

License

MIT

Known limitations (v0.3)

  • Subprocess sandbox only — not production-grade isolation
  • Credential stripping is not a network namespace
  • Single-node, no warm pool
  • TypeScript client is source-only (not published to npm)
S
Description
No description provided
Readme MIT
189 KiB
Languages
Python 95.9%
TypeScript 2.6%
Dockerfile 1.5%