Show HN: Atom – "시각적" 에피소드 메모리를 갖춘 오픈 소스 AI 에이전트

Developer Note: For technical setup and architecture, see docs/DEVELOPMENT.md. Automate your workflows by talking to an AI — and let it remember, search, and handle tasks like a real assistant. Atom is an open-source, self-hosted AI agent platform that combines visual workflow builders with intelligent LLM-based agents. Just speak or type your request, and Atom's specialty agents will plan, verify, and execute complex workflows across your entire tech stack. Key Difference: Unlike SaaS alternatives, Atom runs entirely on your infrastructure. Your data never leaves your environment. Comparing alternatives? See Atom vs OpenClaw for a detailed feature comparison. | Aspect | Atom | OpenClaw | |---|---|---| | Best For | Business automation, multi-agent workflows | Personal productivity, messaging workflows | | Agent Model | Multi-agent system with specialty agents | Single-agent runtime | | Governance | ✅ 4-tier maturity (Student → Autonomous) | ❌ No maturity levels | | Memory | ✅ Episodic memory with graduation validation | ✅ Persistent Markdown files | | Integrations | 46+ business (CRM, support, dev tools) | 50+ personal (smart home, media, messaging) | | Architecture | Python + FastAPI + PostgreSQL/SQLite | Node.js + local filesystem | | Setup | Docker Compose (~15-30 min) | Single script (~10-30 min) | # Install Atom pip install atom-os # Initialize atom init # Add your API keys to .env # OPENAI_API_KEY=sk-... # MINIMAX_API_KEY=... (optional, for MiniMax M2.7 support) # Start Atom atom start # Open dashboard open http://localhost:8000 That's it! 🚀 Choose your edition: - Personal Edition - Free, single-user, SQLite (default) - Enterprise Edition - Multi-user, PostgreSQL, monitoring → pip install atom-os[enterprise] - Build complex workflows using just your voice - Natural language understanding — no proprietary syntax - Real-time feedback as Atom visualizes its reasoning - Sales Agent: CRM pipelines, lead scoring, outreach - Marketing Agent: Campaigns, social posting, analytics - Engineering Agent: PR notifications, deployments, incidents - Hive Orchestration: Autonomous multi-agent coordination with Queen (Architectural Design) and King (Execution Orchestration). ✨ NEW - Intelligent request routing: Automatic intent classification to route requests to the most capable specialty agent. ✨ NEW - Autonomous Workflow Realization: Convert natural language requests into persistent, executable blueprints via the Queen Agent. ✨ NEW - Governance: Progress from "Student" to "Autonomous" as they gain trust - Rich interactive presentations (charts, forms, markdown) - Live operation visibility: See exactly what agents are doing - Multi-view orchestration (browser, terminal, canvas) - Smart error resolution with actionable suggestions - Full Details → - Experience-based learning: Agents store and retrieve past experiences using the Reflection Pool. - Recursive Self-Evolution: Autonomous critique-based optimization loop (optimized for MiniMax M2.7). - Dual-Trigger Graduation: Skills progress from SUPERVISED toAUTONOMOUS via post-task event hooks and background audits. - Dynamic Streak Rule: Promotion based on "Clean Run" streaks (Success + No Intervention + High Compliance). - Hybrid storage: PostgreSQL for state + LanceDB for mistake memory. - Agent Graduation Guide → - 4-tier maturity-based routing and approval system - AI-powered training duration estimation - Every action logged, timestamped, and traceable - 46+ business integrations: Slack, Gmail, HubSpot, Salesforce, Zendesk - 9 messaging platforms: Real-time communication - Atom SaaS Sync: Bidirectional sync with Atom marketplace for skills and ratings ✨ NEW - Use /run ,/workflow ,/agents from your favorite chat app - Recursive Knowledge Retrieval: Higher-order reasoning via BFS graph traversal. - Canonical Anchoring: Link graph nodes to concrete database records (Users, Tickets, Formulas). - Bidirectional Sync: Update entity metadata directly from the Graph UI. - Visual Explorer: Interactive D3-powered graph visualization at /graph . - GraphRAG Documentation → - 5,000+ OpenClaw/ClawHub skills: Import community-built tools - Skill Marketplace: PostgreSQL-based with search, ratings, categories - Enterprise security: LLM-powered scanning (21+ malicious patterns) with hazard sandbox - Skills Registry: Import with status tracking (Untrusted → Active → Banned) - Dynamic Skill Loading: importlib-based hot-reload with watchdog monitoring - Skill Composition: DAG workflows with parallel execution and rollback - Auto-Installation: Python + npm with conflict detection and automatic rollback - Python Package Support: NumPy, Pandas, 350K+ PyPI packages with vulnerability scanning - npm Package Support: Lodash, Express, 2M+ npm packages with governance - Supply Chain Security: Typosquatting detection, dependency confusion prevention, postinstall malware blocking - Community Skills Guide → - Browser automation via CDP (scraping, form filling) - Device control (camera, location, n