Skip to content

🌟 Chidori x WisePick - Replayable Routing & Deterministic Agent Semantics #38

Description

@w2jmoe

Hi 👋

Been exploring Chidori v3 recently — really impressed by how seriously the project approaches deterministic execution and replay semantics ✨

Especially:

  • replay-first architecture
  • checkpoint/resume
  • observable side effects
  • deterministic runtime behavior

Those ideas overlap quite naturally with what WisePick has been exploring around capability routing 😊
Repo: https://github.com/w2jmoe/WisePick

One idea we keep returning to is:

routing feedback shouldn’t reset to zero every runtime

Meaning runtimes shouldn’t have to repeatedly rediscover:

  • which capabilities are stable
  • which execution paths create retry pressure
  • which routing decisions have poor ROI

WisePick has recently been focused on:

  • replayable routing
  • bounded determinism
  • routing snapshots
  • execution efficacy feedback

rather than orchestration ownership.

Seeing Chidori’s emphasis on replay and deterministic runtime semantics made me wonder whether routing state itself may eventually become a:

  • replayable
  • inspectable
  • benchmarkable

runtime artifact 👀

Very cool project — starred 🍨

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions