claw/docs/superpowers/plans/2026-04-20-generated-scene-runtime-url-classification-hardening-plan.md

Generated Scene Runtime URL Classification Hardening Plan

Date: 2026-04-20 Status: Draft Parent route:

• runtime_url_classification_hardening Parent ledger:
• tests/fixtures/generated_scene/generated_scene_source_first_runtime_semantics_ledger_2026-04-20.json

Plan Intent

Implement the first reusable slice that separates runtime URL roles during generation.

Fixed Input Bucket

Use the bounded bucket:

1. scenes with strong source evidence for multiple URL roles
2. scenes whose current generated manifest only exposes target_url
3. high-signal browser-script scenes where runtime context URL and module-route URL are likely to diverge

This first slice should focus on the highest-risk parameterized browser families before broader expansion.

Allowed Files

1. src/generated_scene/analyzer.rs
2. src/generated_scene/generator.rs
3. src/generated_scene/ir.rs
4. route-local generator tests

Forbidden Files

1. no callback-host/runtime implementation
2. no service-console changes
3. no direct edits to generated skills
4. no board or validation assets

Expected Coverage Delta

1. generated metadata can distinguish app-entry/runtime-context/module-route roles
2. callers are no longer forced to guess page_url semantics for the bucketed scenes

Stop Statement

Stop after the first reusable URL-classification slice is implemented and route-local follow-up assets are published.

Do not expand to every scene in this route plan.