Skip to content

agentic-proteins

agentic-proteins is the strict compatibility package in bijux-proteomics. It preserves legacy runtime imports and entrypoints long enough for callers to migrate to bijux-proteomics-runtime. Its value is not new capability. Its value is controlled continuity while the old surface is retired without breaking downstream users in the dark.

flowchart LR
    callers["legacy callers<br/>imports, CLI paths, API usage"]
    bridge["agentic-proteins<br/>compatibility forwarding"]
    runtime["bijux-proteomics-runtime<br/>canonical execution"]
    lower["foundation, core, knowledge,<br/>intelligence, lab"]

    callers --> bridge
    bridge --> runtime
    runtime --> lower
    bridge -. retire when callers move .-> runtime

Why This Package Still Matters

  • it absorbs migration pain so the canonical runtime can keep moving
  • it keeps legacy entrypoints visible instead of burying them in silent shims
  • it provides a defensible retirement surface: each preserved path should have a reason to exist, not only inertia

What It Owns

  • compatibility forwarding for legacy runtime imports
  • preserved legacy CLI and API entrypoints while migration is still justified
  • the proof bar for keeping or retiring those preserved surfaces

Start With

  • Open Foundation when you need the package boundary first.
  • Open Interfaces when the question is a preserved import, CLI, API, or compatibility contract.
  • Open bijux-proteomics-runtime as soon as the question becomes current execution behavior rather than legacy forwarding.

What Should Make A Reader Suspicious

  • a feature that is attractive even without any legacy caller
  • business logic growing inside the bridge instead of inside runtime
  • compatibility wording that no longer points to a real migration path

First Proof Check

  • packages/agentic-proteins
  • packages/agentic-proteins/src/agentic_proteins
  • packages/agentic-proteins/tests

Boundary

If the behavior would still be desirable after legacy callers disappear, it probably belongs in the canonical runtime package instead of here.