HN 표시: Navox 에이전트 – HITL 체크포인트가 있는 Claude 코드용 AI 에이전트 8개

A specialist AI engineering team for Claude Code. 7 agents. No platform. No login. Your code never leaves your machine. nom.sh — one prompt, 7 minutes: 🦀 A crab cookie clicker. 1,330 lines. 6 bugs caught by QA. See the code → PipeWar — built, debugged, and deployed by agents: 🛡️ A Factorio-inspired tower defense game. Built from scratch, 8 production bugs diagnosed and fixed, 65 tests passing. All by the agent team. Play PipeWar → · See the code → Navox Agents is available on Claude Code's plugin marketplace. This is the fastest way to get started — no cloning, no copying files. If you hit an SSH error, run this first (one time): git config --global url."https://github.com/".insteadOf "[email protected]:" Then install: /plugin marketplace add https://github.com/navox-labs/agents /plugin install navox-agents /reload-plugins If this saves you time, ⭐ star the repo — it helps others find it. Note: Plugin commands are namespaced. Use /navox-agents:agency-run and/navox-agents:hire-team instead of/agency-run and/hire-team . If you installed via the manual copy method below, no namespace is needed. git clone https://github.com/navox-labs/agents.git mkdir -p ~/.claude/agents ~/.claude/commands ~/.claude/templates cp -r agents/.claude/agents/* ~/.claude/agents/ cp -r agents/.claude/commands/* ~/.claude/commands/ cp -r agents/templates/* ~/.claude/templates/ Open Claude Code in any project folder and run: If you installed as a plugin: /navox-agents:agency-run Build a {browser-based} {Cookie Clicker game} with {Atari pixel art} vibes where {crabs eat cookies}. No authentication. No backend. Single HTML file, runs in any browser. Make it {addictive} and {funny}. If you installed manually (copy method): /agency-run Build a {browser-based} {Cookie Clicker game} with {Atari pixel art} vibes where {crabs eat cookies}. No authentication. No backend. Single HTML file, runs in any browser. Make it {addictive} and {funny}. Replace the {variables} with your own idea. This is exactly how we built nom.sh in 7 minutes. flowchart TD CLI["`**$ /agency-run** your prompt here`"] CLI -->|your prompt| AD subgraph ARCH ["🏗️ Architect · Opus 4.6"] AD["DIAGNOSE\nreads request · maps team · flags blockers"] ADE["DESIGN\nstack · auth · API contracts · security · testing"] AD -->|team plan| ADE end ADE -->|system design doc| G1 G1{{"⚠️ Gate 1\nyou review + approve"}} G1 -->|auth + UX brief| UX G1 -->|auth model + threats| SEC1 subgraph PARALLEL1 ["runs in parallel"] UX["🎨 UI/UX · Sonnet 4.6\nFLOW → WIREFRAME → SPEC\nwireframes · component specs · all states"] SEC1["🔐 Security · Opus 4.6\nDESIGN-REVIEW\nauth model · threat surface · constraints"] end UX -->|wireframes + specs| G2 SEC1 -->|auth constraints| G2 G2{{"⚠️ Gate 2\nyou approve — cleared to build"}} G2 -->|cleared to build| FS1 FS1["⚙️ Full Stack · Sonnet 4.6\nBUILD\ncode + auth implementation + unit tests"] FS1 -->|working code + tests| CP CP{{"⚠️ Checkpoint\napp running locally · LGTM / FEEDBACK / STOP"}} CP -->|code to test| QA CP -->|code to audit| SEC2 subgraph PARALLEL2 ["runs in parallel"] QA["🧪 QA · Sonnet 4.6\nTEST-RUN\n38 tests · auth matrix · edge cases"] SEC2["🔐 Security · Opus 4.6\nCODE-AUDIT\nOWASP · auth bypass · vulns"] end QA -->|bug report| FS2 SEC2 -->|vulnerability report| FS2 FS2["⚙️ Full Stack · Sonnet 4.6\nFIXES\nall Critical findings resolved · clean push"] FS2 -->|all findings resolved| G3 G3{{"⚠️ Gate 3\nSecurity LAUNCH-AUDIT · verdict"}} G3 -->|APPROVED ✓| DEVOPS DEVOPS["🚀 DevOps · Sonnet 4.6\nDEPLOY\nVercel + Cloudflare · README written"] DEVOPS -->|live URL + video| SHIP SHIP["🚀 SHIP"] | Agent | What they do | | |---|---|---| | 🏗️ | Architect | Designs the system. Picks the stack. Defines auth. | | 🎨 | UI/UX | Maps user flows. Specs every screen and state. | | ⚙️ | Full Stack | Builds it. Tests it. Ships clean code. | | 🚀 | DevOps | CI/CD. Docker. Deploys. Secrets never touch code. | | 👁️ | Local Review | Starts the app. Shows it to you. Waits for your go. | | 🧪 | QA | Finds every bug. Auth flows get extra scrutiny. | | 🔐 | Security | Audits everything. Nothing launches without a verdict. | | 📦 | Installer | Helps you discover and install individual agents. | Use one agent directly: /architect DIAGNOSE , /security LAUNCH-AUDIT , /qa PLAN (Plugin users: prefix with navox-agents: e.g. /navox-agents:architect DIAGNOSE ) Each agent supports multiple modes — specific workflows you trigger by name. | What you need | Command | |---|---| | Don't know where to start | /architect DIAGNOSE | | Full system design | /architect DESIGN | | Build a feature | /fullstack BUILD | | Check if it's secure | /security LAUNCH-AUDIT | | Run the whole team | /agency-run your task here | | See all modes | docs/modes.md | Plugin users: prefix commands with navox-agents: (e.g. /navox-agents:architect DIAGNOSE ) - Agents pause at every gate and wait for your approval - Nothing destructive runs without your explicit sign-off - You can redirect, reject, or stop at any point Agents stop. They wait. You decide. Then they continue. Full guide: docs/hitl.md First time using Navox Agents on a new project? Tell the agents your stack once — they'll know it every session. After installing the agents globally, run this from inside your project folder: cp ~/.claude/templates/nextjs.CLAUDE.md ./CLAUDE.md # Next.js + Vercel cp ~/.claude/templates/node-api.CLAUDE.md ./CLAUDE.md # Node.js + Express cp ~/.claude/templates/rails.CLAUDE.md ./CLAUDE.md # Rails 8 cp ~/.claude/templates/python-fastapi.CLAUDE.md ./CLAUDE.md # Python + FastAPI cp ~/.claude/templates/cloudflare-workers.CLAUDE.md ./CLAUDE.md # Cloudflare Workers Pick one. The agents read it automatically when Claude Code opens. - Not a platform. No dashboard, no login. - Not a SaaS. No subscription, no usage limit. - Not a walled garden. It's a plugin, but the source is open — fork it, customize the prompts, make it yours. - Not storing your data. Everything runs locally through Claude Code. - Not autonomous. You stay in the loop. 📖 Docs · ⚡ Install · 🦀 See it work · 🐛 Report Bug Built by Navox Labs · MIT License