Dependency Governance¶

Dependency governance is really boundary governance under another name.

For bijux-proteomics-intelligence, dependency review should keep evidence, contracts, and lab execution explicit instead of letting the package collapse into a hidden application layer.

Governance Model¶

flowchart TB
    change["new or changed dependency"]
    purpose{"improves evaluation or explanation work?"}
    boundary{"knowledge, contracts, and lab seams stay explicit?"}
    copying{"dependency avoids copied neighbor semantics?"}
    accept["dependency is governable"]

    change --> purpose
    purpose -->|yes| boundary
    purpose -->|no| reject1["reject or relocate"]
    boundary -->|yes| copying
    boundary -->|no| reject2["reject or isolate"]
    copying -->|yes| accept
    copying -->|no| reject3["redesign the integration"]

This page should make it obvious that a recommendation package gets weaker when it starts owning evidence semantics or execution behavior through dependency shortcuts.

Review Rules¶

guard the seams to evidence, contracts, and lab execution carefully
avoid dependencies that turn the package into a hidden application layer
prefer explicit inputs from neighbors over copied semantics

First Proof Check¶

packages/bijux-proteomics-intelligence/tests
src/bijux_proteomics_intelligence/policies.py and evaluators.py
src/bijux_proteomics_intelligence/report/ and outcomes.py

Design Pressure¶

The easy mistake is to accept a useful dependency that lets intelligence behave like the application shell instead of a bounded decision layer.