Skills
98,067 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Implements authentication using Rails 8 built-in generator. Use when setting up user authentication, login/logout, session management, password reset flows, or securing controllers. WHEN NOT: Authorization and permissions (use Pundit policies), API token authentication, OAuth/SSO integration, or role-based access control.
Activation-aware weight quantization for 4-bit LLM compression with 3x speedup and minimal accuracy loss. Use when deploying large models (7B-70B) on limited GPU memory, when you need faster inference than GPTQ with better accuracy preservation, or for instruction-tuned and multimodal models. MLSys 2024 Best Paper Award winner.
Use when setting up code signing, managing certificates, configuring provisioning profiles, debugging signing errors, setting up CI/CD signing, or preparing distribution builds. Covers certificate lifecycle, automatic vs manual signing, entitlements, fastlane match, Keychain management, and App Store distribution signing.
Use when code signing fails during build, archive, or upload — certificate not found, provisioning profile mismatch, errSecInternalComponent in CI, ITMS-90035 invalid signature, ambiguous identity, entitlement mismatch. Covers certificate, profile, keychain, entitlement, and archive signing diagnostics.
Provides guidance for fine-tuning large language models with Axolotl, covering YAML training configs, LoRA and QLoRA, preference training with DPO, KTO, ORPO and GRPO, multimodal models, FSDP and multi-GPU setups, sequence (context) parallelism, NCCL bandwidth tests, dataset formats, compressed model saving for vLLM and llmcompressor, and custom integrations. Use when writing or debugging an Axolotl YAML config, choosing a dataset format for a fine-tuning run, setting up FSDP or context_parallel_size across GPUs, or running DPO, KTO, ORPO or GRPO training. Use when saving a compressed model for vLLM inference, or when writing a custom Axolotl plugin or integration. Not for inference serving or general Hugging Face Trainer scripts outside Axolotl.
Better Auth — framework-agnostic TypeScript authentication & authorization library. Covers setup, email/password, social OAuth (40+ providers), passkeys, magic links, 2FA, organizations, sessions, plugins, admin, hooks, and security hardening. Use when implementing auth with Better Auth: configuring auth instances, adding providers, setting up database adapters (Prisma, Drizzle, PostgreSQL, MySQL, SQLite, MongoDB), integrating with frameworks (Next.js, Nuxt, SvelteKit, Astro, Hono, Express, Elysia, Fastify, Expo), managing sessions, or extending with plugins.
Core rules for bkit — PDCA methodology, level detection, agent triggering, quality standards, Sprint management (8-phase container with 4 auto-pause triggers, v2.1.13), and Trust Level scope (L0-L4 gates PDCA + Sprint auto-run). Triggers: bkit rules, core rules, methodology
Get the team's Requirements-level backlog (Product Backlog Items and Bugs) filtered to Active items assigned to the current user. Shows parent/child hierarchy and sorts by priority. Use this skill when the user wants to see their backlog items, sprint assignments, or work they own on a team board.
Read-only brainstorming mode for exploring ideas, plans, architecture, debugging approaches, product decisions, research topics, codebase questions, and alternatives without mutating the user's codebase. Use when the user invokes $brainstorm, says "brainstorm", "brainstorming session", "let's think", "explore this", "research only", "read-only", or otherwise asks to discuss before implementation. Remain read-only until the user explicitly asks to implement, edit, write, apply, commit, install, or otherwise change state.
Business Analyst (/ba, alias: Anna, /anna) - Senior BA with 10+ years bridging business and technical teams. Use when conducting market research, competitive analysis, gathering/refining requirements with behavioral acceptance criteria, creating business process models, cost-benefit analysis, financial modeling, user research, or validating assumptions with data. Works closely with /po to shape requirements.
Writing and executing C# scripts and macros against Power BI semantic models using Tabular Editor 2/3. Automatically invoke when the user mentions "C# script", "Tabular Editor script", "TOM scripting", "MacroActions.json", "XMLA", or asks to "automate model changes", "bulk update measures", "create calculation groups", "write a macro", "format DAX expressions", "manage model metadata".
Check and fix accessibility (a11y) on front-end projects (web and mobile web), including Next.js, React, Vue, Angular. Use when the user asks about accessibility, a11y, WCAG, screen readers, voice control, Voice View, keyboard navigation, focus management, ARIA, semantic HTML, color contrast, or fixing accessibility issues in HTML, React, Next.js, Vue, or other front-end code. For native mobile apps (React Native, iOS, Android), see reference; patterns differ.
Check ABAP code for compliance with Clean ABAP principles. Use this skill when users ask to check, validate, review, or analyze ABAP code for clean code compliance, code quality, best practices, or adherence to Clean ABAP guidelines. Triggers include requests like "check this ABAP code", "is this clean ABAP", "review my ABAP for clean code", "validate ABAP against clean code principles", or "analyze ABAP code quality".
Structure software around the Dependency Rule: source code dependencies point inward from frameworks to use cases to entities. Use when the user mentions "architecture layers", "dependency rule", "ports and adapters (hexagonal)", "onion architecture", "screaming architecture", "where should business logic go", "decouple from the database", "swap the framework without a rewrite", or "keep business rules independent". Also trigger when deciding which layer code belongs in, isolating core logic from infrastructure, defining module boundaries, or debating whether the framework should call your code or the reverse. Covers component principles, boundaries, and SOLID. For code-level quality, see clean-code. For domain modeling, see domain-driven-design.
Apply a disciplined engineering workflow to any code change. Use whenever implementing a feature, fixing a bug, or refactoring — before writing code, not after. Walks orient → baseline → smallest change → test → verify → self-review, and enforces language-agnostic hard gates (don't mass-reformat, keep the linter and type-checker clean, keep the build and tests green, make interface changes additive, protect security invariants).
Analyze and improve a software project's module boundaries, dependency direction, file organization, code duplication, oversized modules, and unnecessary indirection. Use when a repository has become hard to navigate, files are proliferating, responsibilities are scattered, or a safe architecture/refactoring plan is needed. Supports analysis-only recommendations and incremental implementation across languages and frameworks.
Generate a DeepWiki-style structured wiki for any codebase to accelerate developer onboarding. Use this skill whenever a user wants to understand, document, or onboard into an unfamiliar repository — including requests like 'explain this codebase', 'generate documentation for this repo', 'help me understand this project', 'create an onboarding guide', 'map out the architecture', or 'how does this codebase work'. Also trigger when a user provides a repo URL or file tree and asks for analysis, or when onboarding new team members into existing codebases. Even if the user doesn't say 'wiki' or 'onboarding', trigger this skill if they want a structured understanding of a codebase.
Use this skill to spawn specialized OpenAI Codex CLI subagents for code review, debugging, architecture analysis, security audits, refactoring, documentation, comparative evidence adjudication, and autonomous /goal runs. It adds each persona through process-local developer instructions while preserving the project's AGENTS.md chain, including during parallel launches. Supports GPT-6-Astra across low, medium, high, xhigh, max, and ultra reasoning with default or priority service tiers, plus GPT-5.6 and GPT-5.5 models. Triggers on 'delegate to Codex', 'Codex Astra', 'Astra subagent', 'Codex subagent', 'code review agent', 'security audit', 'refactor with Codex', 'goal run', 'autonomous goal', 'have Codex weigh this'.
Use when building data integration pipelines, querying industrial data, deploying data models, managing access control, or developing applications that connect to industrial assets, time series, events, and files. Reach for this skill when working with REST APIs, SDKs (Python/JavaScript), data modeling, extractors, transformations, or infrastructure-as-code deployments. Use for **Cognite Flows** custom web apps hosted in CDF (React, `@cognite/cli`, Aura design-system UI with `@cognite/aura`, app hosting, local dev with Vite, deploy, and certification). Core data modelling (CDM)—use when working with CogniteAsset, CogniteTimeSeries, CogniteFile—the CDM concepts for assets, time series, and files. Reach for this skill when building asset hierarchies, populating time series, creating and linking files, or when searching for "Asset" in the API reference (CDM assets are in the Instances API. The legacy /assets endpoint is deprecated as of 2026-10-15 and retires on December 31, 2027)
Compound Engineering workflow for AI-assisted development. Use when planning features, executing work, reviewing code, or codifying learnings. Follows the Plan → Work → Review → Compound loop where each unit of engineering makes subsequent work easier. Triggers on: plan this feature, implement this, review this code, compound learnings, create implementation plan, systematic development.
Anthropic's method for training harmless AI through self-improvement. Two-phase approach: supervised learning with self-critique and revision, then RLAIF (RL from AI Feedback). Use for safety alignment and reducing harmful outputs without human labels. Powers Claude's safety system
Formats commit messages using the Conventional Commits specification. Use when creating commits, writing commit messages, or when the user mentions commits, git commits, or commit messages. Ensures commits follow the standard format for automated tooling, changelog generation, and semantic versioning.
Garnish — Craft CMS's built-in JavaScript UI toolkit for the control panel. ALWAYS load when writing, editing, or reviewing JavaScript that runs in the Craft CP — plugin CP assets, custom field type JS, element index JS, CP webpack config, or code importing garnishjs / referencing window.Garnish. Also for CP accessibility, keyboard interactions, drag-sort, and CP JS memory issues.
Covers DeepSpeed for distributed deep learning training and I/O: ZeRO optimization stages, pipeline parallelism, FP16/BF16/FP8 training, 1-bit Adam, sparse attention, and DeepNVMe (aio_handle, gds_handle, async_io and gds operators, ds_nvme_tune, ds_report) for fast transfers between NVMe storage and host or GPU tensors. Use when configuring ZeRO stages for large-model training, enabling mixed precision or 1-bit Adam, offloading parameters or optimizer state to NVMe with ZeRO-Infinity, writing or reading tensors to files with blocking or non-blocking DeepNVMe calls, or tuning NVMe I/O settings such as block size, queue depth and parallelism. Not for plain PyTorch DistributedDataParallel without DeepSpeed.
deps.dev API v3 package metadata lookup. Covers version discovery, default/latest version, deprecation and project-status flags, resolved requirements/dependency graphs, URL encoding. Use when querying package versions, checking latest releases, detecting deprecated versions, or looking up dependency metadata via the deps.dev API. Keywords: deps.dev, api.deps.dev, package versions.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.