app-modeling: integrate 2026-07-15 evaluation fixes

[willdavsmith]

📊 Application Graph Diff

Comparing main → willdavsmith-integrate-app-modeling-fixes

✅ No application graph changes detected. The application model is identical between main and willdavsmith-integrate-app-modeling-fixes.

graph TD
    classDef added fill:#dafbe1,stroke:#1a7f37,stroke-width:2px,color:#1a7f37
    classDef removed fill:#ffebe9,stroke:#cf222e,stroke-width:2px,color:#cf222e
    classDef modified fill:#fff8c5,stroke:#bf8700,stroke-width:2px,color:#9a6700
    classDef unchanged fill:#f6f8fa,stroke:#d1d9e0,stroke-width:1px,color:#656d76
    skills["skills\ncontainers"]:::unchanged

Loading

Summary

Integrates the sound, non-duplicative app-modeling guidance developed in #14, #15, #16, #17, #18, #19, #20, and #21.

This is an integration candidate for combined evaluation. It does not replace the focused review history in those source PRs.

Reason for change

The evaluation fixes overlap and occasionally conflict. This branch establishes one fail-closed contract for exact target schemas and Recipes, managed versus developer-supplied secrets, immutable source builds, Recipe-mapped outputs, runtime semantics/readiness, and target Environment availability. It also adds the LiteLLM-proven rule that optional repository-wide dependencies must not become resources unless the selected startup/configuration path requires them.

The current-main behavior of always creating or updating .radius/bicepconfig.json, preserving compatible settings, and using a parent config only as input remains intact.

How to test

• Resolve all relative Markdown links under skills/app-modeling.
• Run git diff --check.
• Dispatch app-modeling-eval.yml with skill_repo=radius-project/skills and skill_ref=willdavsmith-integrate-app-modeling-fixes.

File change summary

File	Summary of change
skills/app-modeling/SKILL.md	Consolidates exact-contract precedence, fail-closed behavior, secret handling, source-build constraints, Recipe validation, readiness, and mandatory-only dependencies.
skills/app-modeling/references/architecture-patterns.md	Prevents pattern examples and optional AI/database components from becoming inferred requirements.
skills/app-modeling/references/bicep-structure-rules.md	Aligns generated structure with direct secure inputs, nested managed secrets, exact Recipe outputs, immutable tags, and compatible platforms.
skills/app-modeling/references/connection-conventions.md	Separates native developer credentials from managed-secret projection and rejects connection-only fallbacks.
skills/app-modeling/references/naming-conventions.md	Preserves explicit provider resource-name parameters.
skills/app-modeling/references/runtime-contract.md	Expands evidence and consistency checks for parser coercion, TLS/ports/auth/bootstrap, URL encoding, build metadata, and Recipe availability.
skills/app-modeling/references/secrets-handling.md	Removes authored wrappers for developer credentials and rejects authored-secret URL composition.
skills/app-modeling/references/todo-list-app-example.md	Updates the example for direct secure inputs, immutable tags, compatible platforms, and target Recipe availability.