hxm

Use when the task involves web or mobile accessibility — WCAG audits, ARIA, keyboard navigation, colour contrast, screen-reader behaviour, or making an interface inclusive for every user.

Use when the user asks to bump, release or tag a new version of the current project. Analyses conventional commits since the last tag, proposes a major, minor or patch increment, updates the project manifest and creates the matching git tag once the user has confirmed.

Use when the user asks for a changelog, release notes, or a summary of what shipped since a given tag. Groups conventional commits between two git refs by type and writes the result to stdout or to a file.

Use when reviewing code for security, performance, maintainability, design patterns or technical debt — on a git diff, before a commit, or as a quality gate.

Use when the user asks to commit staged or pending changes. Stages only what the arguments allow, generates a conventional commit message and commits after explicit confirmation.

Use when the user asks to split pending work into several atomic commits, or to commit a large working tree mixing unrelated changes. Groups changes by logical unit and creates one conventional commit per group after explicit confirmation.

Use when working with an external library, framework or package and up-to-date documentation is needed — API usage, version-specific behaviour, configuration or migration.

Use when the user asks to open, create or draft a pull request for the current branch. Detects the base branch, writes an English ticket-based title and body, fills the repository PR template and pushes before creating the PR through the gh CLI.

Use when the user asks to document code, a feature, an API or a system architecture, or to write or refresh a README or technical documentation.

Use when the user asks how long a ticket will take, or for a complexity or effort estimate on a ClickUp ticket. Reads the ticket, maps it onto the codebase and returns a reasoned estimate with its risks.

Use when the user asks to explain, break down or walk through existing code, a pattern, a module or a system architecture, without modifying anything.

Use when a Laravel or Pest test suite fails and the user asks to diagnose or repair it. Classifies every failure as infrastructure noise or as a real regression introduced by the branch, and fixes only the regressions.

Use when the user brings a ClickUp comment or a GitHub review comment and asks to handle, address or answer it. Checks the feedback against the current branch, implements it when founded, and drafts a reply for the user to post themselves.

Use when the user asks for a handoff, for a prompt aimed at another repository, or wants to pass the current work to an agent working in a different codebase. Produces a copy-paste-ready briefing.

Use when the task involves performance — slow pages, Core Web Vitals, database query optimisation, bundle size, server response time or scalability.

Use when the user asks to plan, prepare or scope a ticket before implementing it. Reads the ticket, explores the codebase, resolves ambiguities with the user and produces a step-by-step implementation plan.

Use when the user asks to rebase the current branch, bring it up to date with its base, or resolve rebase conflicts. Auto-resolves trivial conflicts, stops on business-logic ones and force-pushes only with --force-with-lease.

Use when the user asks for a code review of staged changes, of a specific file, or of a diff against a base branch. Reports findings by severity without modifying the code.

Use when the task involves defensive security — vulnerability assessment, OWASP Top 10 compliance, dependency scanning, secret detection or security architecture review.

Use when the user asks for the full specification of a ticket, or to check what a ticket really requires against the codebase. Extracts the requirements, checks the Figma design when one is referenced and lists the gaps.

Use when the user asks to wrap up, summarise or review the work done on the current branch before opening a pull request. Retrieves the linked ClickUp ticket, analyses every change and reports on completeness and quality.