SandBase Harness

What is SandBase Harness?

SandBase Harness is a local-first runtime for AI agents that runs on your machine or your own infrastructure. It covers what sits around the model loop in production: persistent sessions with resumable event streams, sandboxed tool execution, memory stores, credential vaults, permission policies with approvals, and audit and replay, with a built-in Console for people to inspect what an agent did. Sandboxes can be a local process, a per-session Docker container, Kubernetes through kubectl, or a self-hosted worker queue. It supports OpenAI, Anthropic, MiniMax and any OpenAI-compatible provider, including DeepSeek, and exposes an API in the style of Claude Managed Agents so the official Anthropic TypeScript SDK can drive it. Metadata lives in SQLite by default and no hosted control plane is required. It needs Node.js 22 or newer and is free; you only pay your model provider.

Use cases

• Run AI agents in sandboxes you control
• Audit and replay agent sessions
• Store credentials for agent tools in vaults
• Require approval before an agent uses a tool
• Self-host an agent runtime on your own machine
• Run agents on a schedule with deployments

Who it's for

• AI platform engineers
• Backend developers
• DevOps teams
• Security-minded teams

Why use SandBase Harness

• Data stays on your machine or infrastructure, with SQLite and file storage and no required hosted control plane.
• Every session keeps an event log you can replay, which helps when you need to debug or audit an agent.
• Tool access is controlled through permission policies, approvals and credential vaults rather than hard-coded keys.
• You can swap model providers, since OpenAI-compatible endpoints are supported.
• The runtime is free to use; costs come only from your model provider and any infrastructure you add.

SandBase Harness features

• Claude Managed Agents-style /v1 API
• Built-in Console for sessions, settings and API reference
• Resumable Server-Sent Events for replay and debugging
• Sandbox backends: local, Docker, Kubernetes, self-hosted worker
• MCP toolsets, permission policies and built-in tools
• Credential vaults with rotation and OAuth refresh
• Memory stores with version audit
• Skill packages
• Webhook subscriptions and scheduled deployments
• TypeScript SDK

SandBase Harness integrations

• OpenAI
• Anthropic
• MiniMax
• DeepSeek
• Docker
• Kubernetes
• GitHub

How to get started with SandBase Harness

1. Clone the tagged release, then run npm ci and npm run build with Node.js 22 or newer.
2. Run the init command in a new directory to create an agent, a skills folder and config.yaml, then start the runtime.
3. Open Settings > Setup in the Console, paste your provider API key, restart, and set a valid model ID.
4. Create a session for your agent in the Console, or send a message from the chat command.
5. Choose a sandbox backend, review the audit trail and replay events when you need to inspect a run.

SandBase Harness limitations

• Requires Node.js 22+ and npm 10+, plus a model provider API key
• Install only from the tagged GitHub release; the unscoped managed-agents npm package is not this project
• Some Claude Managed Agents routes are partial or unsupported, and each is listed with a reason
• A saved provider setting takes effect only after the runtime restarts

SandBase Harness pricing

SandBase Harness is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

SandBase Harness alternatives

The main SandBase Harness alternatives listed on Debuly are Claude Managed Agents, LangGraph, OpenHands, E2B, Modal and CrewAI. The table compares pricing and platforms; order is A–Z.

1. 

   Claude Managed Agents

   Anthropic's hosted agent harness for long-running, asynchronous tasks: define an agent, pick a sandbox, start a session and stream results.

   Developer platformsAutomation and agentsCoding
   PaidVisit Claude Managed Agents (official site)
2. 

   CrewAI

   Multi-agent framework and platform for building AI agent crews in Python or a visual editor, then deploying and monitoring them.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit CrewAI (official site)
3. 

   E2B

   Cloud sandboxes where AI agents run code safely, with Python and JavaScript SDKs and per-second usage billing.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit E2B (official site)
4. 

   LangGraph

   Low-level open-source orchestration framework and runtime for building stateful, long-running AI agents as graphs.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit LangGraph (official site)
5. 

   Modal

   Serverless cloud platform for running AI inference, training, batch jobs and agent sandboxes, billed per second of compute.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Modal (official site)
6. 

   OpenHands

   Open-source AI coding agent platform with a free Agent Canvas, a Teams plan, and self-hosted Enterprise control

   CodingAutomation and agentsDeveloper platforms
   FreemiumVisit OpenHands (official site)

SandBase Harness FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from avatars.githubusercontent.com · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.