# Hector — Self-Hosted Agent Runtime Binary

> Run Hector as a self-hosted agent runtime: a single Go binary that serves an agent API and studio UI, keeping execution on your own infrastructure.

## Install

Copy the content below into your project:

# Hector — Self-Hosted Agent Runtime Binary

> Run Hector as a self-hosted agent runtime: a single Go binary that serves an agent API and studio UI, keeping execution on your own infrastructure.

## Quick Use

1. Install:
   ```bash
   curl -fsSL https://gohector.dev/install.sh | sh
   ```
2. Run:
   ```bash
   hector serve
   ```
3. Verify:
   - Open the studio UI and confirm you can run a trivial agent test end-to-end.


---

## Intro

Run Hector as a self-hosted agent runtime: a single Go binary that serves an agent API and studio UI, keeping execution on your own infrastructure.

- **Best for:** teams that want to run agents in their own infra with a simple install and a built-in UI for testing
- **Works with:** Single binary install, Docker, and agent runtime patterns; integrates with MCP (per docs)
- **Setup time:** 9 minutes


### Quantitative Notes

- GitHub stars + forks (verified): see Source & Thanks
- Docs advertise a single-binary install + `hector serve` entrypoint (docs)
- Setup time ~9 minutes (install + first run + open UI)


---

## Practical Notes

The fastest evaluation loop is: install → `hector serve` → open Studio → run a toy agent prompt. Then add one real tool (MCP) and validate: permissions, logs, and failure modes. If you can’t explain the security boundary, you’re not ready to put it in front of production data.

**Safety note:** Self-hosting shifts responsibility to you. Define boundaries: network egress, secrets handling, and audit logs before onboarding real repos.

### FAQ

**Q: Is it only for local dev?**
A: No. It targets self-hosted deployments, including Docker-based runs.

**Q: Do I need to be “all in” to try it?**
A: No. Start with one agent and one tool integration, then decide if it fits.

**Q: How does MCP fit in?**
A: Use MCP as the tool layer; keep runtime policies and hosting in Hector.

---

## Source & Thanks

> GitHub: https://github.com/verikod/hector
> Owner avatar: https://avatars.githubusercontent.com/u/194164380?v=4
> License (SPDX): MIT
> GitHub stars (verified via `api.github.com/repos/verikod/hector`): 54
> GitHub forks (verified via `api.github.com/repos/verikod/hector`): 2


---

<!-- ZH -->

# Hector——自托管的 Agent 运行时（一体二进制）

> Hector 是自托管的 agent 运行时：一个 Go 单文件二进制提供 agent API 与 Studio 界面，支持本地或 Docker 部署，让执行留在你自己的基础设施上并便于审计。

## 快速使用

1. 安装：
   ```bash
   curl -fsSL https://gohector.dev/install.sh | sh
   ```
2. 运行：
   ```bash
   hector serve
   ```
3. 验证：
   - Open the studio UI and confirm you can run a trivial agent test end-to-end.


---

## 简介

Hector 是自托管的 agent 运行时：一个 Go 单文件二进制提供 agent API 与 Studio 界面，支持本地或 Docker 部署，让执行留在你自己的基础设施上并便于审计。

- **适合谁（Best for）:** 希望把 agent 运行在自家基础设施、并且想要简单安装 + 内置 UI 测试面板的团队
- **兼容工具（Works with）:** 单文件二进制安装、Docker 部署与 agent runtime 模式；可与 MCP 配合（文档说明）
- **安装时间（Setup time）:** 9 分钟


### 量化信息

- GitHub stars + forks（已核验）：见「来源与感谢」
- 文档强调单文件安装与 `hector serve` 启动方式（文档）
- 跑通约 9 分钟（安装 + 首次启动 + 打开 UI）


---

## 实战要点

最快的验证闭环：安装 → `hector serve` → 打开 Studio → 跑一个玩具 agent 测试；然后接入一个真实工具（MCP），验证权限、日志与失败模式。如果你说不清安全边界，就不要让它接触生产数据。

**安全提示：** 自托管意味着责任在你：在接入真实仓库前先定义边界——网络出站、密钥管理与审计日志。

### FAQ

**Q: 只能本地开发用吗？**
A: 不止。它面向自托管部署，也支持 Docker 运行。

**Q: 试用需要大迁移吗？**
A: 不需要。先从 1 个 agent + 1 个工具集成开始，确认契合度再决定是否深入。

**Q: MCP 在里面怎么用？**
A: 把 MCP 当工具层，运行时策略与托管交给 Hector 管。

---

## 来源与感谢

> GitHub：https://github.com/verikod/hector
> Owner avatar：https://avatars.githubusercontent.com/u/194164380?v=4
> 许可证（SPDX）：MIT
> GitHub stars（已通过 `api.github.com/repos/verikod/hector` 核验）：54
> GitHub forks（已通过 `api.github.com/repos/verikod/hector` 核验）：2


---
Source: https://tokrepo.com/en/workflows/hector-self-hosted-agent-runtime-binary
Author: Agent Toolkit