Separate workflow, perspective and entry point

These examples and illustrative results are independently authored teaching materials, not measured model results.

Use case

A release entrypoint needs code/security reports. Teaching leaf roles inspect domains, workflows supply steps and the entrypoint owns collection/decision scope. A single security question does not require nested coordinator personas.

Mechanism

Separate process, perspective and entrypoint using minimal routing. Directly invoke one lens; fan out independent lenses and collect/verify/synthesize centrally. Avoid layers repeating the same task. Reports do not independently trigger publication.

Bad example

Each checker creates another coordinator, repeatedly paraphrases/delegates and assumes someone else collects; every role may decide to publish.

Good example

The teaching entrypoint gets two leaf reports at one revision/input/format. Roles use relevant checking workflows without redundant coordination layers. The entrypoint collects/verifies with disagreement and actual state retained. Publication follows its separate scope; single-lens work is direct with explicit ownership.

Why the change matters

Clear responsibilities converge on an entrypoint with full task scope. Avoiding useless relays preserves identity and prevents ambiguous collection ownership.

Observable expectation

Illustrative flow is entrypoint→code/security→entrypoint verification. One owner synthesizes, and missing output stays incomplete. Inspect no leaf publishing or recursive relay.

Two roles existing proves no executed review.

Limits

Flat reports fit known artifacts; causal investigation may need interaction. Source persona restrictions are workflow choices, not a universal ban on useful hierarchy. Every coordination layer needs a real responsibility.