zby/llm-do
View on GitHub ↗ ★ 21 Python
"Extend with llm, stabilise with code" - Prompts when you're exploring, Python when you're certain. Workers and tools form one function space—refactor between them.
4 mentions on Hacker News by 2 people: 0 stories, 4 comments · first 2025-11-25 · latest 2026-07-15
20082026
-
…does not offer a unified namespace for functions and prompts (agents) like https://github.com/zby/llm-do (currently in suspended development). There are now also the 'dynamic workflows' in Claude Code that are pretty interesting appr
comment by zby · 2026-07-15 · in Towards a harness that can do anything -
Here is my theory about weaving deterministic code and prompts: https://github.com/zby/llm-do/blob/main/docs/theory.md . Plus a library that realises the unified call space that I propose. I think co-recursion between prompts and code is…
comment by zby · 2026-03-01 · in Deterministic Programming with LLMs -
…should be symbolic and where it should be llm based (fuzzy/neural). This is the idea that started https://github.com/zby/llm-do - there is also RLM: https://alexzhang13.github.io/blog/2025/rlm/ RLM is simpler - but my approach also have…
comment by zby · 2026-02-01 · in Towards a science of scaling agent systems: When and why agent systems work -
That’s why we need progressive hardening: https://github.com/zby/llm-do/blob/main/docs/concept_spec.md
comment by zbyforgotp · 2025-11-25 · in The Bitter Lesson of LLM Extensions