Action Preflight: consequence-aware admission for LLM agent actionsgithub.com1 ポイント·投稿者 gfernandf1·9 日前·2 コメント
Claude Leak Confirms It: LLM Systems Are Architecture, Not Prompts (Orca)github.com2 ポイント·投稿者 gfernandf1·2 か月前·2 コメント
gfernandf1·2 か月前·議論One thing I didn’t make clear in the post:The core issue is that prompt-based agents recompute reasoning every time.ORCA tries to make reasoning reusable — more like software, less like a one-shot prompt.Working on a minimal example to show this clearly.
gfernandf1·2 か月前·議論you can also download the paper from zenodohttps://zenodo.org/records/19438943slightly different version but the core ideas remain the same
gfernandf1·3 か月前·議論If helpful, I can also share the PDF directly or specific sections (design principles, execution model, etc.)
https://github.com/gfernandf/agent-skills