Browse Skills

Influence And Negotiation helps draft responses and plan strategy for situations that need another person's agreement, including sales, salary asks, hard 1:1s, mediation, and diplomatic pushback. Use it when someone says “what do I say?” or when a deal, raise, or decision is stuck.

NextAuth.js Authentication implements NextAuth.js v5 with Google OAuth, credentials login, JWT sessions, protected routes, and RBAC. Use it when building sign-in flows, session handling, or route protection in Next.js apps.

Product Diagnosis uses a six-question framework to validate product direction before development starts. Use it to assess problem fit, user needs, scope, architecture alignment, and risk for G0 proposals.

Progressive Disclosure Design explains when to keep a Claude skill in one SKILL.md and when to move content into references. Use it when creating or refactoring skills, or when deciding how to keep large skills reliable and context-efficient.

Session Management saves, resumes, and compacts Claude Code sessions when work spans multiple context windows. Use it when context is getting full, switching tasks, or continuing a long-running task later.

Sprint Plan SOP loads sprint planning, postmortem, and architecture context before assessing task complexity and deciding whether to enter Plan Mode. It is used to decompose work and prepare task assignments.

Strategic Compact helps you compact Claude Code sessions without losing important decisions, progress, or debugging state. Use it when context usage reaches about 60-80% to save state and preserve what matters.

Artist Self-Evolve distills prompt patterns and art rules from a completed game project so future projects can reproduce comparable assets. Use it when the lead dispatches a game slug during the self-evolve flow to build an art pack and contract notes.

Discipline defines a failure-prevention protocol for Claude Code sessions. Use it before work to avoid fabrication, missed constraints, weak verification, and sloppy retries.

Optimize creates VectorBT parameter-search scripts for trading strategies. It tests combinations, plots heatmaps, and compares the best results against a NIFTY benchmark.

RalphInit creates current_PRD.md and current_PROGRESS.md from a system plan for an agent-driven development loop. Use it when starting a project that will be developed incrementally across sessions.

Sanity Check asks an agent to summarize what is being built, why, progress so far, and whether the approach still makes sense. Use it for plain-language status checks and course correction.