From 441ba296ccaff5b518ac6e8410a447c98b05a57f Mon Sep 17 00:00:00 2001 From: root Date: Mon, 31 Aug 2026 06:27:30 +0800 Subject: [PATCH] feat: add SandBase Harness --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 937af86..c49799c 100644 --- a/README.md +++ b/README.md @@ -37,6 +37,7 @@ OS-level isolation uses Linux namespaces, cgroups, or kernel-enforced policies; - [yolobox](https://github.com/finbarr/yolobox) - CLI running AI agents in containers where home directory stays unmounted for safety - [CodeRunner](https://github.com/instavm/coderunner) - Secure local sandbox to run LLM-generated code on macOS using Apple containers and MCP - [SandboxAI](https://github.com/substratusai/sandboxai) - Docker-containerized runtime for safely executing AI-generated Python and shell commands +- [SandBase Harness](https://github.com/sandbaseai/sandbase-harness) - Local-first TypeScript agent runtime with governed MCP tools, persistent sessions, approvals, audit/replay, and Local, Docker, Kubernetes, or worker execution backends; isolation depends on the selected backend and deployment - [llm-sandbox](https://github.com/vndee/llm-sandbox) - Python library executing LLM-generated code in Docker/Kubernetes/Podman with customizable security - [MCP Runner](https://github.com/abir-taheer/mcp-runner) - MCP server platform using Docker with gVisor runtime for multi-tenant isolation - [PythonSafeEval](https://github.com/s3131212/PythonSafeEval) - Python execution using nsjail process jail inside Docker containers