al-code-review

Reviews AL/Business Central code diffs at slice and feature gate boundaries with parallel review lenses.

4|Updated Feb 22, 2026
One-click install
npx skills add https://github.com/FBakkensen/bc-agentic-dev-tools-marketplace --skill al-code-review-fbakkensen
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: al-code-review
Source: https://github.com/FBakkensen/bc-agentic-dev-tools-marketplace/tree/main/plugins/al-agentic-dev/skills/al-code-review
Command: npx skills add https://github.com/FBakkensen/bc-agentic-dev-tools-marketplace --skill al-code-review-fbakkensen

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? AL/Business Central changes that pass linters can still carry structural defects, BC anti-patterns, misleading names, and accidental AppSource public-surface lock-in. This Skill runs a deliberate review gate at slice-done and feature-done boundaries so those defects are caught before user verification or merge. ## Core Features & Use Cases - Multi-lens review pass: Spawns parallel review lenses covering project compliance and naming, bug scanning, BC-specific anti-patterns via bc-code-intelligence, code comments and git history, and AppSource public-surface additions. - Adversarial judging and cross-family veto: Deduplicates findings, adversarially tests each one, and runs a second-opinion cross-family pass before anything is reported as must-fix. - Report-only or --fix modes: Reports a must-fix queue, nits, rigor notes, and escalations by default; with --fix it lands fixable findings via a red-green subagent and re-reviews once. - Use Case: After the last technical task of a user-facing slice flips to done, run the review to validate the diff against ADRs, the test specification, and BC vocabulary, then stamp review: clean and open the verify task. ## Quick Start Ask the assistant to run /al-code-review on the current slice diff once all its technical tasks are done and the build is green.

Frequently Asked Questions about al-code-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a code review on an AL/Business Central slice?

Run /al-code-review once every technical task in the slice is done and the build is green. It spawns parallel review lenses over the slice diff, judges findings adversarially, and reports a must-fix queue, nits, and a gate decision.

What does the --fix flag do in an AL code review?

The --fix flag lands must-fix findings in the working tree after the review pass, using a red-green subagent for substantive changes and direct commits for hygiene fixes. It then re-reviews the updated diff exactly once and reports the result.

When should code review run relative to user verification?

Code review runs before user verification, at the slice-done boundary when the last technical task lands. A clean review stamps review: clean and flips the verify task from blocked to ready so the verification track can proceed.

Does this review catch AppSource public API changes?

Yes, a dedicated per-feature lens checks new public procedures, public table fields, and page actions on shipped objects. Linters fire on removal and rename but not addition, so the lens judges whether each addition is intentional lock-in.

What happens when a review finding needs a design decision?

Findings classified as needs-a-decision, such as a new seam or changed behavior, are escalated to /al-steer instead of being fixed. The --fix mode never invents decisions, and cross-family-refuted findings also route to escalation.