All authors
Claude Skills by Gastonchevarria
github.com/Gastonchevarria96 skills1 installs19 views
- Ai Product ArchitectDesigns production LLM system architecture: model routing, streaming and latency, semantic caching, RAG, schema-checked output, guardrails and evals. Use when the user says /ai-arch, 'architect my LLM app', 'design a RAG pipeline', 'cut token costs' or 'which model should I use'.Votes: 0GitHub stars: 3
- Anti SlopRejects and cleans AI-generated slop, low-evidence patterns, lazy type assertions, and performance-degrading anti-patterns in TypeScript and JavaScript. Triggered by /anti-slop, clean code audits, or when refactoring lazy AI code.Votes: 0GitHub stars: 3
- ArchifyBuilds validated architecture, workflow, sequence, data-flow and state diagrams as interactive HTML from plain language, Mermaid or repo code. Use when the user says 'diagram my architecture', 'draw this API sequence', 'map our data pipeline' or 'beautify this Mermaid'.Votes: 0GitHub stars: 3
- Auto CompactDetects context bloat in a long session and proposes compacting it with a structured summary before quality drops. Use when the user says 'the session feels heavy', 'compact', 'I'm running out of tokens' or 'start fresh', or after 30+ tool calls or a major topic change.Votes: 0GitHub stars: 3
- Doubt Driven DevelopmentSends each non-trivial decision to a fresh-context adversarial reviewer before it stands. Use when the user says 'doubt this', 'challenge this decision' or 'verify before we commit', or when stakes are high (production, security, irreversible changes) or the code is unfamiliar.Votes: 0GitHub stars: 3
- GodRoutes a request to the right god-mode skills and subagents and runs it end to end under the Fable 5.1 Protocol. Use when the user types /god or asks for 'full autonomous mode' or 'handle this end to end'. Not for ordinary single-skill requests; use that skill directly.Votes: 0GitHub stars: 3
- Launch Growth LoopBuilds product-led growth into the product: landing pages, activation events, referral loops and lifecycle triggers. Use when the user says /growth, 'add a referral loop', 'define our activation event' or 'PLG funnel'. Not for recurring marketing workflows (use marketing-loops).Votes: 0GitHub stars: 3
- Mvp Scope KillerCuts premature over-engineering from an MVP plan (microservices, custom RBAC, in-house dashboards, hand-rolled auth, speculative abstractions) into a lean core. Use when the user says /scope-kill, 'cut scope', 'is this overkill for an MVP' or 'what can we drop'.Votes: 0GitHub stars: 3
- Saas Business ModelChooses the SaaS business and monetization model: value metric, tiers, usage metering and unit economics. Use when the user says /monetize, 'how should we charge' or 'per seat or usage-based'. For price points use pricing; for billing code use saas-launch-revenue.Votes: 0GitHub stars: 3
- Security And HardeningSecure-coding practices while building features that handle input, auth, sessions or data. Use when the user says 'add login', 'handle file uploads', 'secure this endpoint', 'store user data safely' or 'add a webhook'. For a pre-launch audit use security.Votes: 0GitHub stars: 3
- SecurityPre-launch security audit of the whole app (secrets, injection, dependencies, headers) with a SECURITY_AUDIT.md report. Use when the user says /security, 'audit before launch' or 'check for leaked keys'. For secure coding while building use security-and-hardening.Votes: 0GitHub stars: 3
- Skill InstallerInstalls Agent Skills from a GitHub repo, subfolder or SKILL.md URL, or from a VoltAgent/awesome-agent-skills search, after explicit confirmation. Use when the user says /install-skill, 'install this skill', 'add the skill from this repo' or 'find a Stripe skill'.Votes: 0GitHub stars: 3
- Thermo Nuclear Code Quality ReviewStrict maintainability audit of a branch diff: abstractions, files over 1k lines, spaghetti conditionals; usually launched by thermos. Use when the user says 'thermonuclear code quality review' or 'harsh maintainability review'. For bugs and security use thermo-nuclear-review.Votes: 0GitHub stars: 3
- Thermo Nuclear ReviewStrict audit of a branch diff for bugs, breaking changes, security, devex regressions and feature-gate leaks; usually launched by thermos. Use when the user says 'thermonuclear review' or 'will this PR break anything'. For maintainability use thermo-nuclear-code-quality-review.Votes: 0GitHub stars: 3
- ThermosDouble pre-merge review of a branch or PR diff: bugs and security plus code quality, via two parallel thermo-nuclear subagents. Use when the user says /thermos, 'double thermo review' or 'audit this branch before merge'. Not for a whole repo (use deep-code-audit).Votes: 0GitHub stars: 3
- Agent Mcp BuilderBuilds Model Context Protocol (MCP) servers, tools, resources, and prompts with strict JSON schema validation, stdio/SSE transports, error boundaries, and seamless integration with AI coding agents.Votes: 0GitHub stars: 3
- Ai Product Eval OpsEstablishes continuous evaluation pipelines, ground-truth benchmark datasets, hallucination detection metrics, latency/cost profiling, and regression testing for LLM-powered applications.Votes: 0GitHub stars: 3
- AutoplanRuns the full planning chain in one pass: CEO, design and eng reviews, then spec and task breakdown, merged into one implementation_plan.md. Use when asked to \"plan this end to end\", \"full plan for this feature\", /autoplan or /goal. For one lens, use that plan-*-review.Votes: 0GitHub stars: 3
- Backend ArchitectGuides scalable backend architecture, database schema design, caching topologies (Redis), message queues (Celery/BullMQ/Kafka), webhook delivery systems, and rate limiting strategies.Votes: 0GitHub stars: 3
- Browser Testing With DevtoolsTests and debugs pages in a real Chrome via the chrome-devtools MCP server: live DOM, console errors, network requests, performance traces, screenshots. Use when asked to \"check it in the browser\", \"inspect the network call\", \"profile page load\" or \"verify the UI fix\".Votes: 0GitHub stars: 3
- Code Review And QualityReviews one change, file or small PR for correctness, readability, architecture, security and performance, with severity labels. Use when asked to \"review this code\", \"look over my diff\" or \"review this dependency bump\". Pre-merge branch audit: thermos; whole repo: deep-code-audit.Votes: 0GitHub stars: 3
- Deep Code AuditAudits an entire repository, not a diff: bugs, dead code, TODOs, dummy returns, unfinished endpoints, swallowed errors, missing tests, security, performance. Use when asked to \"audit the whole repo\", \"find incomplete features\" or \"deep code audit\". One diff: code-review-and-quality.Votes: 0GitHub stars: 3
- Error HandlingGuides the agent to write robust Python and TypeScript code by applying error handling best practices, timeouts, and graceful shutdown patterns. Trigger this implicitly whenever writing code that interacts with external services, files, or network requests.Votes: 0GitHub stars: 3
- Fastapi ProGuides development of high-performance asynchronous FastAPI applications with Pydantic v2 schemas, dependency injection, structured error handling, background tasks, and production OpenAPI documentation.Votes: 0GitHub stars: 3
- Founder Technical DecisionDocuments high-impact architecture decision records (ADRs) with explicit categorization of reversibility (Type 1 one-way doors vs. Type 2 two-way doors), trade-offs, opportunity costs, and validation criteria. Triggered by /adr, /decision, or when evaluating critical technical choices.Votes: 0GitHub stars: 3
- Frontend Ui EngineeringBuilds production-quality, accessible, responsive UI components and pages that follow the design system instead of a generic AI look. Use when asked to \"build this page\", \"create a component\", \"make it responsive\", \"fix keyboard navigation\" or \"add loading and empty states\".Votes: 0GitHub stars: 3
- Idea RefineTurns a raw idea into a sharp concept: 5-8 variations, 2-3 stress-tested directions and a one-pager with MVP scope and a Not Doing list. Use when asked to \"help me refine this idea\", \"ideate on\", \"give me variations\" or \"explore other angles\". To validate demand use office-hours.Votes: 0GitHub stars: 3
- Interview MeInterviews the user one question at a time, each with a guessed answer, until it can predict their replies, then confirms a restated intent. Use for vague asks (\"build me X\", no who or why) or when told \"interview me\", \"grill me\" or \"are we sure?\". Runs before idea-refine.Votes: 0GitHub stars: 3
- Nextjs App RouterImplements modern Next.js App Router patterns, React Server Components (RSC), Server Actions, Suspense streaming, parallel/intercepting routes, caching strategies, and optimistic client UI updates.Votes: 0GitHub stars: 3
- Observability And InstrumentationAdds structured logs, metrics, traces and alerts so production behavior can be diagnosed, starting from the questions on-call will ask. Use when asked to \"add logging\", \"instrument this endpoint\", \"set up alerts\", \"add tracing\" or \"we could not tell what happened in prod\".Votes: 0GitHub stars: 3
- Office HoursYC-style test of whether a raw idea is worth building: narrow ICP, hair-on-fire problem, why now, value metric, 48-hour smoke test; verdict proceed, pivot or kill. Use when asked \"is this worth building?\", \"validate this idea\", \"who is the customer?\" or \"office hours\".Votes: 0GitHub stars: 3
- Plan Ceo ReviewReviews a drafted plan, PRD or roadmap as a founder-CEO: KPI impact, 80/20 scope cuts, pricing tier, moat, build vs buy; verdict approve, reduce or reject. Use when asked for a \"CEO review\", \"cut this scope\", \"how does this make money?\" or \"build or buy?\". Raw idea: office-hours.Votes: 0GitHub stars: 3
- Plan Design ReviewReviews planned screens and flows before frontend code: tokens, typography, spacing, micro-interactions, empty states, WCAG contrast, no generic AI look. Use when asked for a \"design review\", \"review this UX\" or \"plan the onboarding screens\". Building the UI: frontend-ui-engineering.Votes: 0GitHub stars: 3
- Plan Devex ReviewReviews an API, SDK, CLI or library from the consuming developer's side: time to first success, typing, actionable errors, copy-paste quickstart. Use when asked for a \"DevEx review\", \"is this API easy to use?\", \"improve these error messages\" or \"review the SDK ergonomics\".Votes: 0GitHub stars: 3
- Plan Eng ReviewStress-tests a technical design before coding: data integrity, failure modes and retries, concurrency, rate limits, test plan, with a risk matrix and verdict. Use when asked for an \"eng review\", \"what could break?\", \"review this architecture\" or \"check failure modes\".Votes: 0GitHub stars: 3
- Planning And Task BreakdownBreaks an agreed spec into small, ordered vertical-slice tasks with acceptance criteria and checkpoints, saved under tasks/. Use when asked to \"break this down\", \"make a task list\", \"what order do we build this?\" or \"split for parallel agents\". Runs no reviews (see autoplan).Votes: 0GitHub stars: 3
- Prompt Engineering PatternsMaster prompt engineering strategies including chain-of-thought, few-shot demonstration calibration, strict JSON schema output constraints, system prompt modularity, and context window optimization.Votes: 0GitHub stars: 3
- Rag ImplementationDesigns and implements advanced Retrieval-Augmented Generation (RAG) pipelines, semantic/hybrid search (vector + BM25), dense reranking, contextual chunking, metadata filtering, and hallucination reduction.Votes: 0GitHub stars: 3
- Ralph LoopExecutes autonomous task loops using disk state, backpressure, and isolated contexts based on the Ralph Loop methodology. Use when the user asks to run a Ralph loop, autonomous task loop, or implement features from specs autonomously.Votes: 0GitHub stars: 3
- RetroConducts post-mortem analysis and retrospective evaluations after shipping a feature, sprint, or resolving an incident. Identifies bottlenecks, extracts reusable patterns, and feeds learnings into persistent agent memory.Votes: 0GitHub stars: 3
- Self Learning SkillsCaptures golden path procedures, reusable workflows, setup steps, and troubleshooting patterns into persisted SKILL.md and AGENTS.md files for continuous agent self-improvement.Votes: 0GitHub stars: 3
- Ab TestingDesigns A/B tests and experiment programs: hypotheses, sample size, metrics and result analysis. Use when the user says 'A/B test this', 'how long should I run this test', 'is this significant' or 'ICE score our backlog'. Not for page conversion audits (use cro).Votes: 0GitHub stars: 3
- Ad CreativeGenerates and iterates ad creative at scale: headlines, primary text, static and video ad concepts within platform specs. Use when the user says 'write me some ads', 'RSA headlines', 'I need more ad variations' or 'iterate on our winners'. Not for targeting or budgets (use ads).Votes: 0GitHub stars: 3
- AdsPlans and optimizes paid campaigns on Google, Meta, LinkedIn and X: structure, targeting, bidding, budgets, retargeting. Use when the user says 'should I run ads', 'my ROAS dropped', 'negative keywords' or 'when should I kill an ad'. Not for bulk ad copy (use ad-creative).Votes: 0GitHub stars: 3
- Ai SeoGets content cited in AI answers (ChatGPT, Perplexity, Gemini, AI Overviews): audits, extractable structure, authority, llms.txt, bot access. Use when the user says 'AEO', 'GEO', 'show up in AI answers' or 'optimize for ChatGPT'. Not for technical SEO audits (use seo-audit).Votes: 0GitHub stars: 3
- AnalyticsSets up and audits marketing tracking: tracking plans, event naming, GA4, GTM, UTMs and debugging. Use when the user says 'set up tracking', 'track conversions', 'UTM parameters' or 'are my events firing'. Not for attribution models or conflicting numbers (use attribution).Votes: 0GitHub stars: 3
- AsoAudits App Store and Google Play listings, scoring metadata, keywords, visuals and ratings into a prioritized action plan. Use when the user shares an app store URL or says 'ASO audit', 'optimize my app listing' or 'why aren't people downloading my app'.Votes: 0GitHub stars: 3
- AttributionMeasures which marketing drives revenue: attribution models, MMM, incrementality, self-reported data, reconciling dashboards, first-party stitching. Use when the user says 'my dashboards disagree', 'first-touch vs last-touch' or 'holdout test'. Not for UTM setup (use analytics).Votes: 0GitHub stars: 3
- Churn PreventionReduces voluntary and involuntary churn: cancel flows, exit surveys, save offers, health scores, dunning. Use when the user says 'people keep canceling', 'build a cancel flow', 'failed payment recovery' or 'churn is too high'. Not for win-back email sequences (use emails).Votes: 0GitHub stars: 3
- Co MarketingFinds and scores co-marketing partners and plans joint campaigns, outreach and agreements. Use when the user says 'co-marketing', 'who should we partner with', 'joint campaign', 'cross-promotion' or 'co-brand'. Not for customer referral programs (use referrals).Votes: 0GitHub stars: 3