Dev open-sources orchestration system that lets local 27B models pick stacks and ship production code
Single_Land8080 · reddit · 2026-08-17
A developer with 8 years of experience open-sourced ai-blueprint-core, an orchestration system encoding architectural know-how into AI. The pain point: AI writes code well but doesn't know what to build — backend, frontend, and database choices are each separate decisions.
The system first picks the stack from project needs: an e-commerce request auto-selects Nuxt.js (SSR for SEO) + .NET 8 backend + MSSQL + Vercel/Azure, then runs backend, frontend, QA, and DevOps agents in sequence to deliver production-ready code. The key selling point is running entirely on local models (Qwen 27B, Gemma 12B) with no expensive API calls — the goal being less tokens, less context, quality software. The author invites the community to find where it breaks.
More from coding & agent
- Codex Tip: Switching reasoning levels invalidates context cache — mark_k · 2026-08-17
- Mediocre Agent Data Science Output? Needs Custom Skills — daveholtz · 2026-08-17
- Running Qwen 3.8 27B on RTX 5060 Ti: 73K Context, Built API with 3 Prompts — chiribe · 2026-08-17
- AI Agents Fail to Respect Field-Specific Norms in Academic Research — daveholtz · 2026-08-17
- OKF vs wikillm: How to choose for agent context? — frakc · 2026-08-17
- Open-source H3 Prompt Studio: local LLM writes MiniMax H3 storyboards — lololerigolo60 · 2026-08-17