Vibe coding a 40k-line codebase with DeepSeek and Kimi K3 audits — what would you change?
whatyathinkk · reddit · 2026-10-13
A Redditor shares an entirely code-never-read vibe coding pipeline and asks for improvements:
- Scale: 40k lines of personal tooling (DB, CLI pipeline, complex GUI), never manually read, still works well.
- Stack: DeepSeek API for daily fixes; local Qwen3.8 Flash Next (Q5, 262k context on 2x5080s + 128GB) for code reviews; occasional €10-20 Kimi K3 full-codebase audits; Hermes as orchestrator with Qwen3.8 27B sub-agents.
- Questions: better model than Kimi K3 for review? Is GLM-5.3 better at security vulns? Worth using Alibaba's open-code-review skill? How to slow codebase rot?
More from coding & agent
- Index Network open-sources Social Discovery Protocol: agents negotiate fit on your behalf — carsonfarmer · 2026-10-23
- Installed a formal-verification skill: it silently installed Java, MacBook runs 10 degrees hotter — zealcaiden · 2026-10-23
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- Ruff author warns: upgrade Node/Bun now — --use-system-ca still accepts the eMudhra cert — charliermarsh · 2026-10-23
- Proof of Agent: paste one prompt and your AI mines, grades, and self-governs — adamamcbride · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23