The safe runtime layer for enterprise AI agents.
SandBase helps teams move agents from demos to production with runtime infrastructure for sessions, tools, approvals, sandboxed execution, memory, audit trails, replay, and operational visibility.
Website · Docs · Blog · Discord · X · LinkedIn
Use the open-source SandBase CLI to connect Codex, Claude Code, Cursor, Windsurf, Gemini CLI, OpenCode, and other AI clients to one MCP bridge. Star the project or install the latest release.
📣 Latest open-source updates: SandBase CLI v0.1.17 connects 25 AI client targets to 2,000+ models and APIs · DSH Plugin Store v0.1.0 Preview 5 installs runtime-verified plugins into a local DeepSeek Harness Web profile · SandBase Harness v0.3.4 is active in the official MCP Registry · SandBase Skills v0.3.4 ships 88 installable Skills as a native DeepSeek Harness bundle
SandBase CLI detects your installed clients and configures the MCP bridge for Codex, Claude Code, Cursor, Windsurf, Gemini CLI, OpenCode, and more:
npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz connectPreview all 25 supported client targets without signing in or changing files:
npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz catalog --jsonSource · v0.1.17 release · Report an issue · Star SandBase CLI
Install all 88 Skills as a native DSH bundle directly from GitHub source:
dsh plugin --profile web add github:sandbaseai/sandbase-skills
dsh webThe flagship multi-source-search Skill works with host-provided search tools
and needs no SandBase account. Need durable managed-agent sessions as native
DSH MCP tools? See SandBase Harness v0.3.4, its
official MCP Registry entry,
and the DeepSeek Harness showcase.
Explore Harness · DSH runtime plugin guide · Independent self-hosting guide · Build an auditable research agent · Star Harness · Browse Skills · Star Skills
Browse more than 4,000 tracked plugin packages from over 3,400 public repositories in a native Store inside DeepSeek Harness:
curl -fL https://github.com/sandbaseai/dsh-plugin-store/releases/download/v0.1.0-preview.5/sandbaseai-dsh-plugin-store-0.1.0-preview.5.tgz -o /tmp/sandbaseai-dsh-plugin-store-0.1.0-preview.5.tgz
dsh plugin --profile web add -w /tmp/sandbaseai-dsh-plugin-store-0.1.0-preview.5.tgzRestart DSH Web, open Settings, and select Store.
Open the project site · View source · Try Preview 5 · Star DSH Plugin Store
When a DSH install, Session, tool call, sandbox, or plugin fails, use the independent DeepSeek Harness Handbook to find the first broken boundary. It currently provides 80 English-canonical, source-backed guides plus local browser tools that do not upload diagnostic input.
Run Install Doctor · Route a failure · Browse all guides · Star the Handbook
Modern agents are no longer just chat interfaces. They call tools, run code, inspect files, browse the web, trigger workflows, and act on behalf of users.
That creates a new infrastructure problem:
- where does generated code run?
- which tools can the agent call?
- who is the agent acting for?
- what happens before a sensitive action runs?
- how do teams observe, replay, and debug agent behavior?
SandBase is built around those runtime questions. The open-source path starts with local-first managed agents; the hosted SandBase platform adds managed sandboxes, team controls, observability, connectors, and production support.
| Area | Focus |
|---|---|
| Agent runtime | Sessions, event logs, replay, memory, and resumable agent execution |
| Tool governance | MCP/tools, permissions, approval patterns, and action policies |
| Sandboxed execution | Safer code, shell, file, browser, and workspace operations |
| Multi-executor support | Runtime patterns that can work across Claude, OpenAI, local models, and custom executors |
| Observability | Logs, audit trails, status, and operational surfaces for agent runs |
| Open resources | Ecosystem maps, cookbooks, labs, and growth playbooks for builders |
| Project | What it is for |
|---|---|
| SandBase Harness | Local-first TypeScript agent runtime with sandboxed execution, MCP integration, approvals, event streams, replay, DeepSeek V4 support, and a containerized MCP bridge. |
| SandBase Skills | 88 installable Agent Skills across Codex, Claude Code, Cursor, and Gemini CLI; the flagship multi-source research workflow runs with host search tools and no SandBase account. |
| SandBase CLI | One-command MCP onboarding for Cursor, Claude Code, Codex, Windsurf, Gemini CLI, OpenCode, and other agents, with access to 2,000+ AI models. |
| managed-agents | Open-source, local-first managed-agent runtime with a Console, Claude Managed Agents-style resource APIs, skills, files, credential vaults, memory stores, environments, and resumable session events. |
| deepseek-harness-handbook | Independent, agent-first field guide to DeepSeek Harness with source-backed quickstarts, architecture maps, safety boundaries, runnable examples, troubleshooting, and multilingual navigation. |
| DSH Plugin Store | Native Community and Installed experience for discovering, filtering, installing, and managing packages from more than 3,400 DeepSeek Harness plugin repositories. |
| awesome-native-agent-platforms | A curated list of infrastructure, runtimes, sandboxes, browsers, model routers, and protocols for building production AI agents. |
| sandbase-lab-sitecheck | "Can AI Get It?", a SandBase-powered website AI personality test where an agent visits a site, scores it, writes feedback, assigns personality tags, and generates a shareable card. |
| awesome-agent-runtime | A 500-project landscape of agent runtimes, sandboxes, browser agents, MCP/tool protocols, memory layers, observability, and compute platforms. |
| agent-sandbox-cookbook | Examples, compatibility checks, and field notes for running AI agent tools across sandboxed runtimes. |
| global-ai-cold-start | A public case study on turning SandBase.ai from an invisible early AI infrastructure product into a searchable, developer-facing trust surface. |
- Building a local-first agent runtime? Start with SandBase Harness.
- Adding research workflows to your agent? Install the no-account
multi-source-searchSkill or browse all SandBase Skills. - Want to connect your coding agent to tools and models? Install SandBase CLI.
- New to agent infrastructure? Read the SandBase blog.
- Comparing native agent platforms and runtimes? Start with Awesome Native Agent Platforms.
- Mapping the broader runtime ecosystem? Browse Awesome Agent Runtime.
- Exploring local-first enterprise agent runtime? Browse Managed Agents.
- Building with DeepSeek Harness? Start with the runtime plugin walkthrough, then use the DeepSeek Harness Handbook for operator guides, architecture, and troubleshooting.
- Building agents that run code or call tools? Read the Agent Sandbox Cookbook.
- Want to see an agent judge a website? Try SandBase Lab Sitecheck.
- Following SandBase updates? Join Discord or follow @SandbaseAI.
We care about infrastructure that helps agents act safely and reliably:
- agent runtime and execution boundaries
- MCP servers, tool protocols, and action schemas
- sandboxed compute for code, shell, browser, and file operations
- model gateways and multi-model routing
- evals, tracing, replay, and observability
- authorization, approvals, and pre-action policy checks
- long-running workflows and distributed execution for agents
If you are building in this direction, we would love to learn from you.