cardinalconseils/claude-starter - Page 4

cks-stack

CKS production stack — 26-layer reference: design tools, build, infrastructure, security, observability, AI/LLM, payments.

deep-research

Multi-hop recursive research — deep investigation, source cross-reference, sub-topic discovery, strategic intelligence report.

github-project-setup

GitHub Project Kanban setup — 6-column board, plugin.json config, phase items, attractor mode, AFK observability.

orchestration

Orchestration patterns — dispatch layer, state machines, saga, circuit breakers, idempotency, dead letter queues, multi-worker.

remotion

Remotion video — programmatic video in React: compositions, sequences, animations, audio, rendering, FFmpeg.

rpi

Research-Plan-Implement sub-cycle within CKS phases — R-P-I sequence, quality gates, artifact handoff, iteration refresh.

skill-creator

Create and improve CKS skills — write SKILL.md, frontmatter, workflows, references, trigger descriptions.

sentry-sdk-setup

Set up Sentry error monitoring, tracing, and session replay in any language or framework. Use when: asked to add Sentry, install Sentry SDK, set up error monitoring, configure error tracking, add observability, or integrate Sentry into a project.

shipping-checklist

Pre-launch readiness checklist for production deployment. Use when: preparing to deploy, promoting to production, running release quality gates, or when the user asks 'are we ready to launch'. Covers code quality, security, performance, accessibility,…

core-behaviors

Non-negotiable agent operating behaviors for vibecoding — surfaces assumptions, manages confusion, pushes back on bad patterns, enforces simplicity, maintains scope discipline, and verifies with evidence. Use when: any agent starts work, quality degrades,…

handoff

Compact the current session into a pickup document for the next Claude session. Use when: "handoff", "picking up tomorrow", "switching sessions", "context for next session", "don't lose context", "continue later", "save state". Also invoke before /compact on…

user-profile

Human user profile — loads the CKS operator's personal preferences, communication style, and domain context from ~/.cks/user-profile.md. Use when personalizing responses to the operator: tone, technical depth, what to optimize for, what to avoid. Trigger on:…

cicd-starter

Bootstrap an existing codebase for the CKS lifecycle — scans the project, runs guided intake, generates CLAUDE.md, and initializes .prd/ state. Use when: "bootstrap", "set up", "initialize", "configure for this project", "generate CLAUDE.md", "scan my…

scheduled-agents

Recurring/scheduled agent patterns — analytics with memory of prior runs, sentiment monitoring with parallel fan-out, and asset generation from repo + brand assets. Use when setting up a CronCreate-based agent that runs autonomously on a cadence.

payments

Stripe payment integration expertise. Use when implementing Stripe Checkout, Payment Intents, Stripe Elements, Stripe Billing, subscription billing, idempotency keys, Stripe webhook handling, refunds, chargebacks, PCI DSS compliance, dunning, invoicing, or…

anti-patterns

Catalog of patterns to detect and reject before they ship. Use when: reviewing code, writing SUMMARY.md, doing code review, verifying sprint output, or any time you are about to claim work is done. Four categories: code, process, output, PRD/doc.

legibility

PMC Legibility Framework — every idea has a legible face (pitch/deck) and an illegible face (truth: gut, relationships, timing, tacit knowledge). Explore Mode gates sprint eligibility. Commit Mode catches plan-vs-reality gaps before building. Use when…

parallel-dispatch

Parallel worker dispatch for the Build node — task grouping rules (disjoint file paths, max 4 groups, domain split), worktree lifecycle, sequential merge strategy

peers

Session awareness and deconfliction via claude-peers-mcp. Shows what all Claude Code sessions in a repo are doing, auto-announces status via hooks, detects conflicts (two sessions on same feature), and enables directives (stop, redirect) from a main session.…

caching

Caching strategy selection and implementation for production applications. Use when: choosing between write-through and write-behind cache, adding Redis or Memcached, reducing write or read latency, designing cache invalidation, implementing read-through or…

database-design

Database schema design, migration discipline, and data modeling for production applications. Use when: designing database tables, creating migrations, choosing between SQL and NoSQL, adding indexes, defining relationships, or reviewing data architecture.

database-recovery

Database backup and recovery strategy selection for production applications. Use when: designing backup schedules, implementing WAL or point-in-time recovery (PITR), choosing between full/differential/incremental backups, defining RTO and RPO targets, or…

performance

Performance optimization and Core Web Vitals for production applications. Use when: optimizing page load, measuring Core Web Vitals, reducing bundle size, fixing N+1 queries, optimizing images, adding caching, running Lighthouse audits, or when users report…

caveman

Compresses agent output into caveman speak — drops articles, filler, hedging, and verbose phrasing while preserving 100% technical accuracy. Use when the user asks for caveman mode, terse output, token reduction, brutal brevity, or activates /cks:caveman.…

ai-marketing

AI marketing expertise — AI-native content strategy, AEO/GEO optimization, llms.txt, prompt-matched copy, AI search citation building, AI directory presence for product/SaaS sites

brand-marketing

Brand marketing expertise — brand voice, identity system, domain authority benchmarking, backlink gap analysis, citation building, brand consistency auditing backed by Ahrefs data

online-marketing

Online marketing expertise — keyword opportunity discovery, content gap analysis, funnel architecture, email sequences, paid ads briefs, CRO — backed by Ahrefs MCP and DataForSEO API for real SERP and traffic data

library-skills

Library-skills integration — installs official AI skills from Python/JS packages into .claude/skills/ for always-current library patterns. Use when: FastAPI detected, Streamlit detected, setting up a Python project, or keeping library skills up to date after…

observability

Application observability and log triage — querying live signals from logs, error tracking, and LLM traces. Use when: debugging production issues, pulling Sentry errors, reading runtime logs, analyzing LangSmith traces, or investigating incidents with live…

ciso

Personal CISO domain expertise — threat intelligence, infrastructure audit, supply chain attacks, RLS policy validation, secrets hygiene, GitHub Actions hardening, webhook exposure, MCP server attack surface. Covers PayFacto/Cardinal Conseils/ServiConnect…

environment-management

Environment separation and configuration management for production applications. Use when: setting up dev/staging/production environments, managing environment variables, configuring secrets, adding feature flags, or deploying across environments.

openrouter

OpenRouter AI gateway for target projects. Use when: the project calls AI APIs, needs model selection advice, wants cost optimization across providers, requires fallback reliability, or is deciding between OpenRouter vs direct frontier APIs. Covers live model…

auto-mode

Claude Code Auto mode permissions — hands-off execution with classifier-based safety. Use when: running long autonomous tasks, replacing --dangerously-skip-permissions, or advising on permission setup. Pair with Plan mode: Plan for scoping, Auto for execution.

ultrareview

Cloud-based deep code review with verified findings only. Use when: reviewing high-stakes code (auth, payments, RLS, voice/AI logic), reviewing a GitHub PR, or when /cks:review noise-to-signal ratio is too high. Do NOT suggest for trivial changes, style…

karpathy-guidelines

Behavioral guardrails derived from Andrej Karpathy's observations on LLM coding pitfalls. Use when: writing, reviewing, or refactoring code; agent is over-engineering; agent is making silent assumptions; agent is modifying code outside its scope; agent is not…

agent-safety

Leash container security for AI agents — Cedar policy generation, minimal-privilege patterns, stack-aware policy templates for sandboxing Claude Code sessions

failure-taxonomy

Classify failures into structured categories and select recovery recipes. Use when: diagnosing errors, sprint failures, QA failures, build errors, test failures, MCP issues, branch conflicts, or any failure that needs classification before repair. Enables…

accessibility

Web accessibility essentials (WCAG 2.1 AA) for production applications. Use when: building UI components, reviewing designs, adding forms, implementing navigation, checking color contrast, or preparing for production launch. Covers the 80/20 of accessibility…

api-design

REST API and interface design conventions for production applications. Use when: designing API endpoints, creating route handlers, defining request/response schemas, handling API errors, adding pagination, versioning APIs, or reviewing API architecture.

authentication

Authentication and authorization patterns for production applications. Use when: adding user login, signup, password reset, social login, role-based access, session management, JWT tokens, OAuth2, MFA, or any auth-related feature. Also use when reviewing auth…