hypeJunction/ai-assistant-starter

finish

End-of-session routine. Ensures test coverage, performs self-review, runs validation, and commits cleanly. Use when finishing a unit of work.

Langue du texte source : anglais

implement

Full feature implementation workflow with explore, plan, code, test, validate, and commit phases. Use for new features, enhancements, or significant code changes.

Langue du texte source : anglais

iterate-pr

Iterate on an open PR until CI passes and all review feedback is addressed. Fetches status, categorizes findings by severity, applies fixes, and loops until clean.

Langue du texte source : anglais

plan

Create a detailed implementation plan without writing code. Read-only analysis and planning with user approval gate. Use before implementing features or making significant changes.

Langue du texte source : anglais

security-review

Systematic security audit with confidence-based reporting. Analyzes attack surfaces, checks against OWASP categories, and reports only confirmed or likely vulnerabilities. Use for pre-merge security review or periodic audits.

Langue du texte source : anglais

validate

Run validation checks to ensure code quality, security, and correctness. Supports quick (scoped), full (CI pipeline), fix (auto-correct), and CI mirror modes.

Langue du texte source : anglais

accessibility-review

Systematic WCAG 2.1 AA accessibility audit with confidence-based reporting. Scans UI components for keyboard, screen reader, and visual accessibility issues. Use for pre-release audits or after major UI changes.

Langue du texte source : anglais

commit

Review changes and create a git commit with user confirmation. Use when work is ready to commit, changes need staging, or the user says "commit".

Langue du texte source : anglais

git-conventions

Git branch naming, commit message conventions (Conventional Commits), workflow patterns, and common operations. Auto-loaded when working with git.

Langue du texte source : anglais

research

Ethical web research with attribution tracking, license compliance, and citation-based output. Use for technical research, competitive analysis, or content synthesis requiring source transparency.

Langue du texte source : anglais

add-story

Create comprehensive Storybook stories with test plans, play functions, and validated tests for a component. Use when a component needs Storybook coverage or story files.

Langue du texte source : anglais

add-todo

Document deferred work, shortcuts, and technical debt for future resolution. Use when taking a shortcut, finding tech debt, or deferring out-of-scope work.

Langue du texte source : anglais

adr

Capture an Architecture Decision Record documenting context, reasoning, alternatives, and consequences of a significant technical decision.

Langue du texte source : anglais

api-test

Discover, design, implement, run, and report on API endpoint tests. Ensures every route is tested for status codes, response shapes, auth boundaries, and edge cases.

Langue du texte source : anglais

debug

Systematic bug investigation and fixing with hypotheses, root cause analysis, regression tests, and verification. Use when encountering bugs, errors, or unexpected behavior.

Langue du texte source : anglais

deps

Audit, update, and manage project dependencies safely. Use when checking for vulnerabilities, updating outdated packages, or performing routine dependency maintenance.

Langue du texte source : anglais

docs

Add or improve documentation for code. Use when code needs JSDoc comments, inline explanations, README files, or documentation updates.

Langue du texte source : anglais

e2e

End-to-end testing workflow for web applications using Playwright or Cypress. Covers user flow discovery, test design, implementation, execution, and flaky test management.

Langue du texte source : anglais

explore

Understand code without making changes. Read-only exploration of codebase structure, patterns, data flow, and dependencies. Use when asked "how does X work" or to investigate code before planning.

Langue du texte source : anglais

hotfix

Emergency bug fix with abbreviated validation for production issues. Use when production is broken, a critical security vulnerability is discovered, data corruption is occurring, or a user-blocking bug has no workaround.

Langue du texte source : anglais

init

Bootstrap AI assistant with project-specific configuration by analyzing the codebase and generating populated template files. Use when setting up a new project or re-initializing after major changes.

Langue du texte source : anglais

migrate

Safe database/schema migration workflow with risk assessment, rollback planning, and ORM-aware generation. Use for schema changes, adding/removing columns, creating tables, adding indexes, or data backfills.

Langue du texte source : anglais

pr

Create a well-documented GitHub pull request with quality checks, proper description, and test plan. Use when pushing a branch, creating a merge request, or preparing code for review.

Langue du texte source : anglais

refactor

Systematic multi-file refactoring with pattern analysis, scope detection, batched execution, and progress tracking. Use for renames, pattern changes, API migrations, or any change affecting 6+ files.

Langue du texte source : anglais

release

Prepare and publish a release with version bump, changelog, and tagging. Use when shipping a new version, creating a release candidate, or tagging a production build.

Langue du texte source : anglais

revert

Safely rollback changes using git revert with impact assessment and validation. Use when a commit needs to be undone, a PR introduced a bug, or changes need to be rolled back without rewriting history.

Langue du texte source : anglais

review

Comprehensive code review of the current branch against base. Read-only analysis with P0-P3 severity-rated findings and actionable feedback. Use before merging or to check code quality.

Langue du texte source : anglais

sync

Audit and align AI documentation with the actual state of the codebase. Use when documentation feels stale, after significant refactoring, periodically at sprint boundaries, or when AI assistance seems to follow outdated patterns.

Langue du texte source : anglais

tdd

Strict Test-Driven Development workflow enforcing the RED-GREEN-REFACTOR cycle. Ensures every line of production code is justified by a failing test. Use when building new features, fixing bugs, or adding behavior.

Langue du texte source : anglais

test-coverage

Analyze branch changes and ensure adequate test coverage. Creates missing tests with test plans, runs them, and reports results. Use after implementing changes to add tests.

Langue du texte source : anglais

track-files

Track batches of files that need similar changes for large-scale work. Use when refactoring affects many files, migrating patterns, or coordinating batch operations across sessions.

Langue du texte source : anglais

branch-protection

Runtime enforcement hook that blocks dangerous git operations on protected branches. Prevents force-push, hard reset, and branch deletion on main/master. Auto-loaded for all git operations.

Langue du texte source : anglais

communication-guidelines

Communication templates and response formatting for AI coding assistants. Defines structured message formats with semantic indicators for status, errors, actions, and progress. Auto-loaded for all interactions.

Langue du texte source : anglais

destructive-command-protection

Runtime enforcement hook that blocks destructive system and database commands. Prevents accidental rm -rf, database drops, and other irreversible operations. Auto-loaded for all command execution.

Langue du texte source : anglais

github-actions-guidelines

CI/CD guidelines for GitHub Actions including pipeline structure, caching, secrets management, deployment strategies, and quality gates. Auto-loaded when working with workflow files.

Langue du texte source : anglais

ai-assistant-protocol

Core execution protocol governing code quality, testing, scope management, and approval gates for AI coding assistants.

Langue du texte source : anglais

code-review-guidelines

Code review best practices including review checklist, comment types, severity levels, feedback patterns, and PR size guidelines. Auto-loaded during code review.

Langue du texte source : anglais

docker-node-guidelines

Docker guidelines for Node.js including Dockerfile best practices, multi-stage builds, Docker Compose, security, and health checks. Auto-loaded when working with Docker configurations.

Langue du texte source : anglais

documentation-guidelines

Documentation best practices for code comments, JSDoc, READMEs, and API docs. Explains when to comment and when not to. Auto-loaded when writing documentation.

Langue du texte source : anglais

env-config-guidelines

Environment configuration guidelines for Node.js including variable naming, type-safe loading with Zod, secrets management, and feature flags. Auto-loaded when working with environment config.

Langue du texte source : anglais