Open-source SafeAI adds 8 more AI capability checks after a 5-day, $8 build
IkarusCareer · reddit · 2026-07-27
Built an open-source static analyzer for AI apps, the author says SafeAI gained its first community contribution and now detects eight more capabilities, including Docker, Kubernetes, Redis, Slack, browser automation, and Google Cloud.
The more interesting part is the build workflow: the project was developed over about five days using multiple models in different roles — GPT Codex 5.3 for architecture and implementation, Kimi K3 for review and refactoring, and DeepSeek V4 for documentation and verification. The author says the mix felt more productive than using a single model, and the whole build reportedly cost about $8 in model usage.
More from coding & agent
- Builder says `/prototype` failed on a teleprompter feature, should have grilled docs first — mattpocockuk · 2026-07-27
- Agent rules backfire again: the model invents a “characterize + defer” excuse — Vjeux · 2026-07-27
- Modified RPG prompt has been running 24 hours, with subagent loops changing what one-shotting can do — mattshumer_ · 2026-07-27
- Claude Code is not reliable enough for long research projects without human supervision — _akpiper · 2026-07-27
- Planr adds a live workflow map to track multiple agents, blockers and dependencies — kevinkern · 2026-07-27
- Long-running agents hit race conditions after immutable event logs — neal_lathia · 2026-07-27