惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Hugging Face - Blog
Hugging Face - Blog
云风的 BLOG
云风的 BLOG
大猫的无限游戏
大猫的无限游戏
M
MIT News - Artificial intelligence
L
LangChain Blog
阮一峰的网络日志
阮一峰的网络日志
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
Recent Announcements
Recent Announcements
IT之家
IT之家
Google DeepMind News
Google DeepMind News
罗磊的独立博客
爱范儿
爱范儿
Last Week in AI
Last Week in AI
人人都是产品经理
人人都是产品经理
U
Unit 42
MongoDB | Blog
MongoDB | Blog
S
SegmentFault 最新的问题
B
Blog
博客园 - 叶小钗
月光博客
月光博客
Stack Overflow Blog
Stack Overflow Blog
V
Visual Studio Blog
C
Check Point Blog
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
I Built a Single-File AI Agent in Go — Zero Dependencies,...
Jason Huang · 2026-05-18 · via DEV Community

Hey DEV community! 👋

I'm an undergrad developer who spent the last few months hacking on something I'm excited to share: OpenAgent — a local AI Agent that ships as a single binary. No Docker. No Node.js. No Python environments. Just download the .exe and double-click.

GitHub: github.com/the-open-agent/openagent


The Problem I Was Trying to Solve

Like many of you, I've been experimenting with AI Agents for everything from code generation to automating tedious tasks. But I kept hitting the same friction:

# Typical "getting started" experience
npm install -g some-agent
# 847 packages installed...
# Oh, you need Python 3.9+ too
pip install ...
# Also Docker for the vector DB
docker-compose up
# 12 minutes later...
# Wait, why is this 400MB?

Enter fullscreen mode Exit fullscreen mode

Sound familiar? I wanted something that felt more like:

# Download openagent.exe (23MB)
# Double-click
# Done ✅

Enter fullscreen mode Exit fullscreen mode

So I built it.


What Makes OpenAgent Different

Single Binary, Zero Runtime Dependencies

I wrote OpenAgent in Go with a specific constraint: everything must compile into one executable. The React frontend gets embedded at build time. The backend is pure Go. One process, one port (14000), one file.

The result:

  • 23 MB executable (vs 400MB+ for typical setups)
  • ~2.7s cold start (vs 30s for alternatives)
  • ~110 MB idle memory

It's Not Just "Hello World"

Despite being a single file, OpenAgent packs real capabilities:

Feature Details
🤖 Models 30+ providers: OpenAI, Claude, DeepSeek, Gemini, Mistral, Grok...
🌐 Browser Real browser automation (navigation, clicks, forms, screenshots)
🖥️ Shell Local command execution with PTY support
📄 Office Read/write Word, Excel, PowerPoint
🔗 MCP Plug-and-play with any MCP-compatible server
📚 RAG PDF/Word/Excel → automatic chunking, embedding, indexing
Workflows BPMN-style visual orchestration
📊 Dashboard Token usage, activity logs, tool management

Why Go Instead of Python?

"But isn't AI/ML Python's domain?"

Here's the thing: Agent performance is bottlenecked by LLM API calls, not language execution speed. The magic happens in the architecture — how you orchestrate tools, manage state, handle concurrency.

Go gives me:

  • Static compilation → single distributable binary
  • Goroutines → efficient concurrency without the memory bloat
  • Cross-platform → Windows/Mac/Linux from one codebase
  • No runtime → users don't install anything

For a "personal local Agent," these trade-offs beat Python's dynamic flexibility.


Quick Start

Windows (easiest):

  1. Download openagent_Windows_x86_64.exe from Releases
  2. Double-click
  3. Open http://localhost:14000

macOS/Linux:

curl -fsSL https://raw.githubusercontent.com/the-open-agent/openagent/master/scripts/install.sh | bash

Enter fullscreen mode Exit fullscreen mode

Configure your model:
The UI has a one-click setup for OpenAI, Claude, DeepSeek, or any OpenRouter-compatible provider.


The Engineering Decisions I'm Proud Of

1. Shell with Guardrails

Shell access is powerful but dangerous. In tool/shell.go, I built in:

  • Default 30s timeout (max 300s)
  • Session-based poll/write/submit flow
  • Optional PTY for interactive programs
  • Audit logging for every command

It's designed as a production tool, not a foot-gun.

2. Memory-Conscious Concurrency

I ran an 80-concurrent health check stress test. Memory grew by 10 MB. Go's goroutine + channel model makes this trivial compared to the memory curves I've seen in Node.js-based agents.

3. Frontend Embedded, Not Served

The React build gets embedded into the binary with Go's embed package. No separate static files to manage, no path resolution issues, no "where did my assets go" bugs.


Real Talk: What's Not Perfect

I want to be transparent about where we stand:

Metric OpenAgent OpenClaw
Lighthouse Score 45 87
Health Check P50 ~33ms ~20ms

Our frontend loading needs work. The Lighthouse gap is real — we're optimizing it. The health check difference is milliseconds, but still.

If you find cases where we're slower or more expensive, file an issue. I genuinely want to know.


Who This Is For

  • Developers who want a Claude Code alternative that runs locally with their own API keys
  • Privacy-conscious users who don't want data leaving their machine
  • Resource-limited setups where running multiple agents used to be impossible
  • Security folks who need audit trails and controlled shell access

Get Involved

OpenAgent is Apache 2.0 licensed. The code is on GitHub, and I review issues/PRs within 12 hours.

If the project helps you, a ⭐ means a lot. If you want to contribute, there are plenty of good first issue labels waiting.

GitHub: github.com/the-open-agent/openagent

Discord: discord.gg/5rPsrAzK7S


The Bigger Picture

I believe personal AI Agents should be lighter, not heavier. One file. Double-click. Done.

No 400MB installs. No dependency hell. No "it works on my machine."

Just an Agent that does the job and gets out of your way.

That's OpenAgent.


Built with Go, caffeine, and the stubborn belief that software should just work.