8 open-source repos to self-host a lean AI agent stack instead of stacking $20/month SaaS
SaiTarunRonanki · reddit · 2026-10-28
A Reddit user lays out a fully open-source, self-hosted 2026 AI agent + automation stack across 8 GitHub repos: Ollama (local model runner), vLLM (high-throughput inference), LangGraph (graph-based agent state machines), Smolagents (1,000-line code-first agents), LlamaIndex (data connectors), MCP reference servers (standard tool protocol), n8n (visual orchestration with human-approval gates), and Browser Use (Playwright-driven web workflows). Their practical takeaway: pair n8n orchestration + MCP tool servers + a cheap/local model router before calling frontier models to keep costs predictable and keep humans in control of any send, pay, or delete action.
Related event: Redditor Builds Self-Hosted AI Agent Stack with 8 Open-Source Projects(2 posts)→
More from coding & agent
- Anthropic models burn 2-3x tokens and cost 10-50x more on FrontierCode 2.0 — teortaxesTex · 2026-10-28
- What to learn in the AI coding era: the bottleneck moved from writing to verifying code — irombie · 2026-10-28
- Devs praise OpenAI's Decisions API: Codex integration drives full-time switch — craigsdennis · 2026-10-28
- Personal AI agent trains a chess transformer on dual RTX 4090 and battles Sunfish — andreisavu · 2026-10-28
- Opus 5.5 ultracode workflow with 9 subagents builds a Three.js scene — nptacek · 2026-10-28
- Agent sandboxes talk at PLATFORMday SF: the missing runtime for AI — gethackteam · 2026-10-28