softspark/ai-toolkit

API Patterns covers REST and GraphQL API design, including resource naming, versioning, pagination, idempotency, OpenAPI, error contracts, and safe retries. Use it when designing or reviewing API endpoints and failure responses.

Accessibility Validator scans a codebase for WCAG 2.1 AA, EN 301 549, and EAA issues without changing files. Use it to check contrast, keyboard support, ARIA, media, forms, and accessibility-statement compliance.

Agent Creator creates new specialized agents with frontmatter, tools, model choice, and delegated skills. Use it when defining a new agent, routing triggers, or validating an agent scaffold.

App Builder scaffolds new projects for web, mobile, API, CLI, and AI apps using stacks like Next.js, Vite, Nuxt, Astro, FastAPI, Django, Laravel, React Native, and Flutter. Use it when starting a new app or choosing a starter architecture.

Architecture Audit explores a codebase for architectural friction, shallow modules, and tight coupling. It surfaces candidates, compares interface designs, and files an RFC issue when you need to deepen a module boundary.

Architecture Decision documents ADR, RFC, or RFD choices with context, constraints, options, trade-offs, recommendation, and consequences. Use it when deciding between architectural approaches or recording why a choice was made.

Biz Scan analyzes a codebase for business opportunities, KPI coverage, monetization gaps, and missing data exposure. Use it when you need an advisory report on analytics instrumentation and revenue-related surfaces.

Brainstorm prices the zero option before any artifact exists and can end with do not build. Use it to explore half-formed ideas, separate the real problem from a proposed solution, and decide whether to stop, route, or proceed.

Brand Voice enforces a direct, concise voice for documentation, README files, and user-facing text. It also supports concise and strict output modes in CLAUDE.md when response length and structure need control.

Briefing generates a decision-focused daily summary from agent logs and session data. Use it to review ops, strategy, and actions that need attention across the system.

Build detects the project toolchain and runs the appropriate build command for Node, Python, Flutter, Go, Rust, or Docker projects. Use it when you need compiled artifacts or packaged outputs.

C# .NET Patterns covers common C# and .NET idioms such as LINQ, async/await, nullable references, records, dependency injection, ASP.NET Core, EF Core, and MediatR. Use it when writing or reviewing modern .NET application code.

C# Rules defines coding style, framework usage, patterns, security, and testing conventions for C#/.NET projects. Use it when writing or reviewing .cs, .csproj, .sln, ASP.NET Core, EF Core, or test code.

C++ Rules covers the project's C++ style, patterns, security, build, and testing standards. Use it when writing or reviewing .cpp, .h, CMakeLists.txt, or GoogleTest code.

CI Pipeline Management generates, updates, or troubleshoots CI/CD configs for projects using GitHub Actions or GitLab CI. Use it to detect existing pipelines, add lint/test/build stages, or debug failing jobs.

CI/CD Patterns provides examples for GitHub Actions, GitLab CI, Docker builds, caching, and Kubernetes rollout strategies. Use it when creating or refining build, test, release, and deployment pipelines.

CVE Scan detects project ecosystems and runs native audit tools to report known dependency vulnerabilities. Use it to check npm, pip, Cargo, Go, Composer, Bundler, or Dart projects and optionally auto-fix supported issues.

Chaos injects controlled faults in non-production to test resilience and recovery. Use it for latency, dependency failure, or other fault-injection experiments with a healthy baseline and recovery checks.

Claude Toolkit Rules sets workspace-wide conventions for Claude Code tasks, including path safety, offering alternatives, and devil's-advocate review. Use it when following the toolkit's planning, workflow, and decision rules.

Clean Code helps you improve readability and maintainability by applying meaningful names, SRP, DRY, small functions, and guard clauses. Use it when reviewing or refactoring code with smells like long methods, god classes, or dead code.

Code Analysis scans a codebase or path for quality, complexity, dependencies, coverage gaps, and security issues. Use it to identify hotspots, patterns, and recommendations without changing source files.

Command Creator creates new Claude Code slash commands with frontmatter, instructions, and validation checks. Use it when you need a new project, user, or plugin command scaffolded to ai-toolkit conventions.

Content Moderation Patterns shows a two-stage moderation pipeline with cheap pre-filters, LLM classification, thresholds, and human review. Use it to enforce content policy with audit trails and structured decisions.

Decision Council evaluates an architecture or product decision from four perspectives using parallel sub-agents and a synthesized recommendation. Use it when comparing options, weighing trade-offs, or checking the impact of a proposed change.

Git Commit creates Conventional Commit messages from staged changes and validates them before committing. Use it when you need a structured git commit with the right type, scope, and safety checks.