bc-agentic-dev-tools-marketplace

Build and test AL/Business Central projects. Use after modifying AL code or tests to verify the build gate passes. Runs compilation, publishing, and test execution in a single command. Required gate before committing AL changes.

AL/Business Central code review at gate points. Report-only by default; pass --fix to land must-fix findings and re-review once. Use when a slice's last technical task lands (slice-done) — both backend-only and user/API-facing slices, before the verify task is refined — when all tasks are done before merge (feature-done), or when the user asks for an in-depth code review.

Pick a `ready-for-implementation` technical task from the `tasks/` folder and drive it red→green through TDD for AL/Business Central. Use after `/al-refine`, one task per session, Unit AAA cases first, then Integration AAA cases. Stops at green and hands off to `/al-refactor` then `/al-mutate`.

Validate AL/Business Central test rigor by mutation: inject one mutation at a time, run the script-backed build gate, classify, revert, report killed/surviving/equivalent mutants. The rigor step the user runs after `/al-refactor` on whatever arrived without a red, or standalone on legacy code before `/al-refactor`.

Reshape AL/Business Central production and test code while tests stay green, via four parallel lens subagents then serial apply with `/al-build` between. Use after `/al-implement` takes a task to green (full task diff, once per task) or standalone on legacy code.

Settle the AL/Business Central feature architecture from idea or `event-model.md`. Use after `/al-event-model` for user/API-facing features, after `/al-grill-adr` for backend-only features, or when the user asks to design an AL feature.

Domain-aware grilling for AL/Business Central. Sharpens BC vocabulary against CONTEXT.md, cross-references intent with the codebase, and offers domain ADRs only when a hard-to-reverse business rule earns one.

One `status: ready` task to a fresh Test Specification or Verification Plan for AL/Business Central. Technical task -> ready-for-implementation. Verify task -> ready-for-verification.

Coach and navigator for AL/Business Central agentic dev. Reads the tasks/ folder, the goal, the codebase, and recent commits, names what is next or blocked or drifting, owns .out-of-scope/ and .not-yet-specified/, and is the canonical replan venue.

Temporarily add `DEBUG-*` `FeatureTelemetry.LogUsage` probes to AL code to inspect runtime state via `.output/TestResults/*/telemetry.jsonl`. Use when runtime behaviour diverges from source and tests alone can't reveal which path ran. Probes are temporary; remove before delivery.

Settle the user-facing journey for AL/Business Central as `event-model.md` in BC vocabulary (Role / Action / Business Event / View / Status). Use after `/al-grill-adr` for user/API-facing features before `/al-design`; backend-only features skip.

Guide the user to record the slice's framework-limited E2E Journey Examples in BC's Page Scripting recorder — one scenario at a time in chat, punchline-first. The user records and downloads the `.yml`; the agent replays each on a fresh container and classifies reds. Recordings are reserved for behaviour no AL test layer can automate (generation-time push-down). Prerequisite to `/al-user-verification`.

Quiz the developer on recently landed AL/Business Central changes — one question at a time, in chat — to keep their mental model in contact with the codebase. Use after a long agentic run, before merging a feature, when returning to a project after time away, or standalone on any diff, slice, or object area the user names.

Verify AL/Business Central specifics from authoritative sources, quote them, return — the evidence-bar escalation seat. Invoke when two sources disagree, when a fact lands in a durable design artifact (event-model.md, architecture.md, CONTEXT.md, ADRs), or when a fuzzy question needs framing plus cross-family verification. Callable from a session and by another skill. Single-fact lookups go direct; this skill arbitrates.

Decompose `architecture.md` into a slice-grouped task list in the `tasks/` folder for AL/Business Central, with one verification task per slice when `event-model.md` is present. Use after `/al-design`, before `/al-provision` on the bracketed `T-001` task.

Guide the user through one slice's `ready-for-verification` verify task in the `tasks/` folder for AL/Business Central — punchline-first, one scenario at a time, in chat. The user walks the non-recorded Journey Examples in their own browser and reports what they see; the agent runs containers, the pre-flight recording batch, and Contract checks, asks one check at a time (ask-before-reveal), records, and routes. Functional outcomes gate; usability observations become findings → tasks.

Execute the `kind: breaking-change` task in the `tasks/` folder for AL/Business Central — run al-build's `validate-breaking-changes.ps1` against the provisioned baseline, then flip the task `done` or `blocked`. Use as the feature's last task; a detected break stops for a human, never self-resolved.

User-facing orientation for the al-agentic-dev plugin — pipeline diagram, 19-skill catalogue plus the subagent prompt blocks, persistence layers, and cold-start guidance. Use when the user asks "what is al-agentic-dev", "what skills are in here", "show me the pipeline", "where do I start from scratch", or wants a tour. Pure static emit; does not inspect repo state. For state-aware navigation mid-feature ("what should I do next"), the dispatcher should prefer /al-steer.

Execute the `kind: provision` task in the `tasks/` folder for AL/Business Central — refresh the build environment (compiler, symbols, and — when enabled — the breaking-change baseline) by running al-build's `provision.ps1`, then flip the task `done` or `blocked`. Use on the first task of a feature, or whenever a `kind: provision` task sits at `status: ready`.

Independent read-only advisory review via cross-family CLI dispatch (shells to GitHub Copilot CLI, pinned to a GPT model). Use from `/al-implement`, `/al-refine`, `/al-refactor`, `/al-user-verification`, or `/al-code-review` before reconciling non-trivial `Test Specification`, `Verification Plan`, mutation lists, refactor checklists, a verification walk verdict, or a review fix-queue.

Generate release notes from per-PR analysis of merged work since the last release. Use after producing .output/releases/release-analysis.jsonl when drafting release notes, summarising changes between releases, or preparing version documentation for an AL/Business Central app.