Skip to content

al-agents-review: derive the target BC version from app.json - #231

Open
Stefano Demiliani (demiliani) wants to merge 1 commit into
microsoft:mainfrom
demiliani:skills/derive-bc-version-from-app-json
Open

Stefano Demiliani (demiliani) wants to merge 1 commit into
microsoft:mainfrom
demiliani:skills/derive-bc-version-from-app-json

Conversation

@demiliani

Copy link
Copy Markdown
Contributor

Why

Almost every review leaf already resolves the target BC version from the consuming app's app.json when the orchestrator doesn't supply one. For example, al-query-review and al-reporting-review say "Use the target version from app.json when available", and al-scm-review uses the application major version from the host context. al-agents-review doesn't. Its Relevance section only says to match "the target BC version", so in a standalone plugin review the version is usually unknown.

That matters more here than anywhere else: all 20 Agents articles are version-gated ([27..] or [28..]). With an unknown version, every one is conditionally applicable, findings are capped at medium confidence, and every message has to name the unknown dimension, even when the app's app.json clearly targets BC27+.

What

One sentence in community/skills/review/al-agents-review.md → Relevance: when no target version is supplied, use the major version of the application field in the consuming app's app.json; otherwise the dimension stays unknown. There is no change to READ semantics, outputs, or other skills.

The same fix is applied to the proposed al-copilot-review in #229.

Validation

.github/scripts/Test-SkillIndex.ps1 and tools/Test-ReviewFixtures.ps1 (including -PrepareDirectory): PASSED.

🤖 Generated with Claude Code

All Agents articles are version-gated ([27..] or [28..]). Without a target
version every one stays conditionally applicable and its findings are
capped at medium confidence. Align the Relevance section with the
Microsoft leaves: when the orchestrator supplies no version, use the
major version of the app.json application field.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@demiliani

Copy link
Copy Markdown
Contributor Author

Hi Jesper Schulz-Wedde (@JesperSchulz), could you take a look at this one? It's a one-sentence change in community/, which has no CODEOWNERS entry, so no reviewer was requested automatically.

Context: almost every review leaf already takes the target BC version from the app's app.json when the orchestrator doesn't pass one (al-query-review, al-reporting-review, al-scm-review, ...). al-agents-review doesn't. All 20 Agents articles are version-gated ([27..]/[28..]), so in a plain plugin review they all stay conditionally applicable: findings are capped at medium confidence and every message has to mention the unknown version, even when the app clearly targets BC27+.

This PR aligns its Relevance section with the other leaves. #229 contains the same fix for the proposed al-copilot-review.

Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant