awesome-ai-coding-sandboxes vs agent-sandbox
Coding-agent sandboxes: directory and local container
Both projects address security boundaries for AI coding agents. awesome-ai-coding-sandboxes is a curated comparison of sandbox options, while agent-sandbox provides a local Docker-based environment with restricted workspace and network access.
awesome-ai-coding-sandboxes: Compare Sandboxes for AI Coding Agents
A security-posture-first directory of sandboxes for running AI coding agents’ code. Compare isolation, outbound network controls, secrets handling, and workspace state to shortlist options for self-hosted or managed deployments.
agent-sandbox: Secure AI Coding Agents with Docker
Agent Sandbox runs AI coding agents in locked-down local containers, using Docker-compatible runtimes and a mitmproxy sidecar to control network access and keep injected secrets out of the agent container. It supports CLI and IDE devcontainer workflows, with Colima and Docker Engine on Apple Silicon as its target setup.
| awesome-ai-coding-sandboxes | agent-sandbox | |
|---|---|---|
| Language | Python | Python |
| License | CC0-1.0 | MIT |
| Stars | 28 | 211 |
| Forks | 16 | 19 |
| Last analyzed | Nov 7, 2026 | Oct 28, 2026 |
Key differences
- awesome-ai-coding-sandboxes compares managed and self-hosted options across microVM, VM, container, process, and WebAssembly approaches; agent-sandbox focuses on running agents in Docker containers.
- awesome-ai-coding-sandboxes organizes its directory around documented isolation, egress controls, secrets handling, and workspace state; agent-sandbox implements workspace restrictions, proxy-based network rules, and proxy-side secret injection.
- awesome-ai-coding-sandboxes is licensed CC0-1.0; agent-sandbox is licensed MIT.
- awesome-ai-coding-sandboxes is a young research directory with no releases and contributions mostly from one person; agent-sandbox is in early-stage development and may introduce breaking changes.
- agent-sandbox supports CLI and devcontainer workflows, with CLI recommended for the tightest boundary; awesome-ai-coding-sandboxes helps teams shortlist tools rather than run agents itself.
Choose awesome-ai-coding-sandboxes if you…
- need to compare sandbox approaches and documented security controls before selecting a tool.
- want to review managed and self-hosted options across different isolation technologies.
- are shortlisting options based on secrets handling, network egress, or workspace state.
Choose agent-sandbox if you…
- want to run an agent locally in a Docker container limited to a project repository.
- need configurable outbound network rules and proxy-side credential injection.
- prefer a CLI or supported devcontainer workflow for agent-assisted coding.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.