Open-source workflow engine seeks real use cases: LLM proposes, deterministic policy approves
RI4D · reddit · 2026-11-13
A developer is building an MIT-licensed workflow engine where the LLM proposes actions but a deterministic policy layer decides what actually runs, aiming to move beyond demos where a human clicks Approve all day.
They're asking on Reddit for real use cases to test against — e.g. replaying failed pipeline events for one customer without touching another customer's data or double-executing effects — and want to know which actions people still keep manual or read-only, and what's missing from current tooling.
More from coding & agent
- Vet dev: AI-written game engines aren't cheating, they're the next Unity moment — bendee983 · 2026-11-13
- Agent politely asks its human to stop messaging: new messages cancel three parallel agents — x5ud0kn1gh7x · 2026-11-13
- Open-source coding agent framework AgenC Core merges 15 fixed issues to main — tetsuoai · 2026-11-13
- Claude Code calls the human user "Claude", sparking identity-confusion laughs — wavefnx · 2026-11-13
- Five coding agents on one local Qwen: Claude Code finishes in 15 min while Codex burns 4 hours and aborts — seti_at_home · 2026-11-13
- Fable 5.2 + Tripo + Blender MCP conjures a skeleton-spawning necromancer from one prompt — sidahuj · 2026-11-13