TanStack
精选
router
TanStack Router is a type-safe, data-driven React router with built-in caching, prefetching, and nested layouts, while TanStack Start extends it into a full-stack SSR framework.
AI 仓库情报摘要
FR-AI / ANALYSIS为什么值得关注
它提供了一个生产就绪的 AI 代理控制平面，本地优先、开源且模型无关，弥补了代理 SDK 与完整运行时基础设施之间的差距，且无供应商锁定。
适合谁使用
- 构建自托管代理平台的企业 AI 团队
- 原型设计生产级代理工作流的开发者
- 需要可审计、沙箱化代理执行且本地数据控制的团队
- 将代理运行时嵌入桌面或内部工具的集成商
典型使用场景
- 运行本地 Claude 托管代理风格控制台进行代理开发
- 构建自托管企业 AI 代理，支持可审计的会话和工具调用
- 原型化客户支持、事件响应或研究代理
- 打包可复用的代理模板、MCP 连接器和技能用于现场部署
项目优势
- 基于 SQLite 的元数据存储，运行时状态位于项目目录之外
- 支持可恢复的服务器发送事件，用于会话时间线、调试、审计和重放
- 内置凭据保险库，支持加密存储和网络范围注入
- 开源（Apache-2.0），提供与 Claude 托管代理兼容的 API 和 SDK
使用前须知
- 需要 Node.js 22+ 和 npm 10+；需要配置模型供应商 API 密钥
- 本地优先设计，仅使用 SQLite；无内置云或分布式存储支持
- 高级沙箱执行（Docker）为可选功能，默认不启用
README 快速开始
managed-agents
SandBase managed-agents is the safe, local-first runtime layer for enterprise AI agents.
managed-agents helps teams move AI agents from demos to production with
runtime infrastructure for sessions, tools, approvals, sandboxed execution,
memory, credential vaults, audit trails, replayable events, and operational
visibility. It exposes a Claude Managed Agents-compatible Console and /v1
resource API while keeping runtime metadata in SQLite outside your project by
default.
Use it to build and operate self-hosted AI agents, local developer agents, desktop agent runtimes, MCP-enabled workflows, and enterprise proofs of concept without locking your runtime layer to a single model provider.
Why managed-agents?
Agent SDKs are great for writing an agent loop. Production agents need more: state, session history, tool governance, sandbox boundaries, credential handling, memory, auditability, and a Console for humans to inspect what happened.
managed-agents focuses on that runtime layer. It is not a visual workflow
builder and it is not another model SDK. It is an open-source control plane for
running, observing, and governing AI agents locally or in self-hosted
environments.
Features
- Claude Managed Agents-style Console and
/v1resource APIs - SQLite-backed agents, skills, sessions, environments, credential vaults, memory stores, API keys, and file metadata
- Resumable Server-Sent Events for session timelines, debugging, audit, and replay
- File resources, memory stores, credential vaults, and environment templates
- local file/skill bytes stored outside the source checkout by default
- One active model provider boundary configured through Settings
- Local API keys and bearer-token authentication for shared local runtimes
- Optional seed/import folders for
agents/*.yamlandskills/*/SKILL.md - Local sandbox execution by default, with advanced sandbox adapters configured only when available
- MCP toolsets, built-in tools, permission policies, and skill packages
- One configured model vendor boundary with adapter-owned model defaults
- Optional TypeScript convenience SDK at
managed-agents/sdk
Common Use Cases
- Run a local Claude Managed Agents-style Console for agent development
- Build self-hosted enterprise AI agents with auditable sessions and tool calls
- Prototype customer support, incident response, researc
项目描述
Open-source CMA-compatible agent runtime. Run multi-agent systems locally with any model (Ollama/vLLM/Claude/GPT), MCP tools, scenario templates, and a beautiful dashboard. One command start. Built for enterprise teams.
相关仓库与替代方案
根据分类、Topic 和编程语言匹配的相似项目。
vercel-labs
精选
scriptc
scriptc compiles ordinary TypeScript into small, fast native executables without needing Node.js, V8, or any JavaScript runtime in the binary.
Jakubantalik
精选
thinking-orbs
A React component library that renders six hand-tuned animated thought orb loading indicators on a plain 2D canvas, with two purpose-tuned sizes and automatic theme detection for AI and agent UIs.