Add read-only AL development plan guidance - #155
Add read-only AL development plan guidance#155Jesper Schulz-Wedde (JesperSchulz) wants to merge 5 commits into
Conversation
Add read-only planning and repository-changing development skills so BCQuality knowledge can guide features, bug fixes, refactors, upgrades, and maintenance before the existing AL review gate runs. Track Microsoft Learn ingestion and add development and BCApps-shaped guidance evaluation fixtures. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 638b66d2-9f06-4f60-8781-808709e1485c
|
👋 Heads up Jesper Schulz-Wedde (@JesperSchulz) — and cc maintainers — this PR introduces new top-level entries that aren't part of BCQuality's known repository structure:
This isn't a block — just a flag. 🚩 New top-level folders and files are usually unintended (a stray export, a tool's scratch dir, or content that meant to land inside an existing layer like If this was intentional and the new entry genuinely belongs at the repo root, a maintainer can review and merge as normal — no action needed beyond a quick sanity check. If it wasn't, please move the content into the right existing layer (or drop it) and push an update. 🙏 A maintainer will take a look before merging. |
Bound post-implementation review rounds, expose output kinds in Entry dispatch, enforce capability coverage, map BCFIX-HANDOFF v1, clarify no-knowledge behavior, and reject repository-escaping skill paths. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 638b66d2-9f06-4f60-8781-808709e1485c
Retain shared knowledge enrichment and review guidance; defer standalone implementation and source-ingestion tracking. Add runner-owned baseline evidence, contract regressions, and explicit consumer/pilot boundaries. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
Reconcile the read-only plan-enrichment contracts with main's folder-review inputs and documentation structure. Record Windows alternate streams in runner evidence and clear the regression harness exit status after expected negative probes. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 638b66d2-9f06-4f60-8781-808709e1485c
Accept main's removal of deterministic compiler and analyzer duplicates, including the ApplicationArea rule, while retaining the read-only development guidance contract and non-mechanical Learn knowledge. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: 638b66d2-9f06-4f60-8781-808709e1485c
Summary
Narrow this PR to the shared, read-only foundation for knowledge-backed AL authoring. Existing development and bug-fix workflows retain ownership of execution.
al-development-planand its host-native adapter: enrich an existing plan and repository context with referenced BCQuality constraints, companion samples, validation considerations, and unresolved applicability.Consumer boundary
The proposed flow is: existing analysis and normalized plan -> read-only BCQuality guidance -> existing implementation phases -> independent final BCQuality review -> delivery.
Consumer-specific plan and continuation-state normalization stays in the consumer. The fixtures use a synthetic metadata-and-Markdown initial plan rather than treating
BCFIX-HANDOFFas the normal planning interface. BCQuality does not own TDD, environment provisioning, retries, state, propagation, critique, commits, or PR creation.no-knowledgemeans no additional applicable BCQuality constraints, not that the change is unsafe or unimplementable. Consumers retain their ordinary gates and own handling of partial, failed, and materially unresolved guidance.Consumers should persist guidance within their own state lifecycle, re-enrich when the plan materially changes, and record/verify the same immutable BCQuality checkout and filtering policy for guidance and final review. This PR does not implement consumer uptake or establish that a configured ref is enforced at runtime.
Evaluation and evidence
These checks establish contract/scorer behavior, not improved generated code. Snapshot comparison is not a sandbox or a monitor of reverted transient writes; it cannot prove that articles were opened or prose constraints are semantically faithful. Real model/runtime evidence remains separate.
Explicit follow-up
Consumer rollout and the authoring-quality pilot are not completed or claimed by this PR.