Skills
97,905 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Build, review, migrate, or debug React interfaces that use Base UI (`@base-ui/react`) unstyled accessible primitives. Use when the user asks about Base UI docs, component anatomy, styling, composition, render props, TypeScript wrapper types, forms, dialogs, drawers, popovers, menus, selects, comboboxes, autocomplete, createItems, tabs, toast, OTP Field, accessibility, or replacing lower-level UI primitives with Base UI.
Build correct Bento integrations and choose the right Bento AI surface across Bento MCP, the Bento CLI, and official Bento SDKs for Node, Laravel, PHP, Drupal, Go, .NET, Elixir, Python, Ruby, and n8n. Use when adding Bento to an application, working with a live Bento account through an AI assistant, syncing subscribers, tracking events, sending transactional email, creating broadcasts, validating email, configuring API credentials, reviewing Bento SDK usage, or choosing the right Bento tool for a project. Emphasizes safe customer account operations, framework-native setup, bulk/upsert-style subscriber imports, secure credentials, queues/jobs/workflows, SDK-specific install commands, and avoiding per-user loops where batch APIs exist.
Use when upgrading from Bevy 0.17 to Bevy 0.18, when an LLM writes `EventReader`/`EventWriter`/`Trigger<E>` instead of `MessageReader`/`MessageWriter`/`On<E>`, when `mesh.insert_attribute` fails to compile, when `AmbientLight` no longer works as a resource, or when `Camera { target: ... }` errors on the `target` field. Index of every breaking 0.17→0.18 change.
Use when upgrading from Bevy 0.18 to Bevy 0.19, fixing resource/query conflicts, replacing `SceneRoot` with `WorldAssetRoot`, migrating text to `FontSource` and `FontSize`, converting render-graph nodes to render systems, replacing lifecycle `Replace` with `Discard`, or updating asset load builders, lights, input focus, and Cargo features.
Lint and format frontend code with Biome 2.4. Covers type-aware linting, GritQL custom rules, domains, import organizer, and migration from ESLint/Prettier. Use when configuring linting rules, formatting code, writing custom lint rules, or setting up CI checks. Triggers on biome, biome config, biome lint, biome format, biome check, biome ci, gritql, migrate from eslint, migrate from prettier, import sorting, code formatting, lint rules, type-aware linting, noFloatingPromises.
File a GitHub issue of type Bug from a description — structures a clear bug report (summary, steps to reproduce, expected vs actual, environment), previews it, then on confirmation creates the issue with the Bug issue type (falling back to a bug label when the repo has no issue types). Use when the user asks to file a bug, open a bug report, create a GitHub bug issue, log a bug, or runs /bug.
Collection of deprecated and breaking change patterns and fix recipes for baserCMS's CakePHP version upgrades (5.0 → 5.1 → 5.2 onward). Covers "Association alias already set", "A validation rule with the name already exists", "_Token not found", "headers already sent", "Plugin named X is already loaded", errors with ResultSet/PaginatedResultSet, deprecation of order()/group(), deprecation of event listener return values, deprecation of find('all', $options), deprecation of array options in Table::get / named arguments, exceptions from null values in array conditions, the _cake_core_ cache configuration, setting.php of subsequent plugins not being read in pluginBootstrap so Configure is null (live iteration), the .. template in _checkFilePath causing MissingPluginException/500, setting parent_id=null on the root in TreeBehavior corrupting lft/rght, deprecation of Entity::set array bulk assignment, AbstractMigration/AbstractSeed deprecation → BaseMigration/BaseSeed (cakephp/migrations 4.5+), fallbacks() in config/routes.php causing 404 only in tests, and a helper's real method being delegated to a magic method so method_exists returns false (TextHelper::truncate → is_callable check), among others. Reference this when investigating and fixing problems caused by CakePHP itself or related packages. When adapting to a new CakePHP version, add a version-specific entry to this document. For problems caused by PHP itself, refer to the php-migration skill.
Build autonomous AI agents with Claude Agent SDK. Structured outputs (v0.1.45, Nov 2025) guarantee JSON schema validation, plugins system, hooks for event-driven workflows. Use when: building coding agents with validated JSON responses, SRE systems, security auditors, or troubleshooting CLI not found, structured output validation, session forking errors.
Use when the user asks to generate or update a project's CLAUDE or AGENTS context file from a codebase scan. Writes a focused file under 100 lines containing only the non-obvious build commands, conventions, and gotchas Claude Code needs.
Create Claude Code plugins, skills, commands, agents, hooks, MCP servers, and marketplace files. Use when building a plugin, packaging skills for sharing, creating a marketplace, converting standalone config to a plugin, or deciding between plugin vs skill vs MCP vs standalone.
Goldsky CLI command and flag reference — all valid subcommands, arguments, and options for goldsky turbo, pipeline, subgraph, secret, project, dataset, indexed, telemetry, and compose. Consult before suggesting any goldsky command to avoid hallucinating invalid commands or flags.
Use when the user asks to explain what a piece of ADVPL/TLPP code does in plain language, adapting the explanation depth for a junior developer (line-by-line), a senior developer (business logic, design decisions, risks), or a functional consultant (no technical terms) -- e.g. understanding legacy code, filling in missing documentation, or assessing the impact of a customization. Also triggers on Portuguese phrasing like "explicar fonte", "explicar codigo", "o que essa rotina faz", "explicar para o consultor funcional", or requests for a --level junior/senior/funcional explanation.
Software-engineering heuristics based on Mark Seemann's Code That Fits in Your Head (2021), updated for agent-driven development. Use when writing or reviewing code, refactoring accidental complexity or a Big Ball of Mud, controlling technical or architectural debt in generated code, designing APIs and invariants, adding a feature through a walking skeleton and acceptance tests, debugging a defect with reproducible tests or bisection, threat-modelling endpoints and trust boundaries with STRIDE, planning a legacy or Strangler migration with rollback, or setting up a maintainable codebase. Covers decomposition and cyclomatic complexity, cohesion, encapsulation, outside-in TDD, separation of concerns, Git/review discipline, safe evolution, and troubleshooting. Not for language syntax, framework tutorials, production incident response, or performance profiling.
Use before any codebase search or exploration in a large or unfamiliar repository — finding where a feature/domain lives, locating a file, deciding which app/package owns some logic — before reaching for find/grep/Explore or a research subagent.
Analyzes project bounded contexts, extracts business rules and domain knowledge, writes ai-context/features/<context>.md files, and produces a teach-report.md with documentation coverage metrics. Trigger: /codebase-teach, teach codebase, extract domain knowledge, update feature docs.
Build a component the way a library author would. Decide primitive versus hand-rolled, cover the full state matrix, design the API by composition, style through data attributes, keep it controlled or uncontrolled deliberately, and pick the right library for the job from a short trusted list instead of installing whatever is first on npm. Use when creating or refactoring a button, input, menu, dialog, drawer, toast, tooltip, table, command palette, tabs, or any reusable piece of UI.
When the user wants to apply computer vision in warehouses, detect defects, track packages, count inventory, or automate visual inspection. Also use when the user mentions "computer vision," "image recognition," "object detection," "barcode reading," "package tracking," "quality inspection," "YOLO," "RCNN," "image classification," or "warehouse automation with vision." For general ML, see ml-supply-chain.
Check and configure feature flag infrastructure using OpenFeature with pluggable providers. Use when adding feature flag infrastructure to a new project, setting up the OpenFeature SDK with a provider (GOFF, flagd, LaunchDarkly), auditing existing feature flag configuration, configuring relay proxy infrastructure, or adding feature flag test helpers. Natural triggers: "set up feature flags", "add openfeature", "configure launchdarkly".
Check and configure code formatting tools (Biome, Prettier, Ruff format, rustfmt). Use when setting up a formatter for a project, migrating from Prettier to Biome or Black to Ruff, auditing formatter configuration for best practices, adding format-on-save and CI format checks, or standardizing formatting across a monorepo. Natural triggers: "set up formatting", "configure biome", "migrate to ruff format", "add prettier".
Fetch up-to-date documentation and code examples for any library, framework, SDK, CLI tool, or cloud service. Use whenever the user asks about a specific library—even well-known ones like React, Next.js, Prisma, Express, Tailwind, Django, or Spring Boot—because training data may not reflect recent API changes or version updates. Always use for API syntax questions, configuration options, version migration issues, “how do I” questions mentioning a library name, debugging involving library-specific behavior, setup instructions, and CLI tool usage. Use even when you think you know the answer. Do not rely on training data for API details, signatures, or configuration options; they are frequently out of date. Prefer this over web search for library documentation.
Use when the user wants to create any new Backstage package — plugin, module, processor, provider, scaffolder action, or library. Drives `yarn new` to scaffold the right template and optionally pre-wires common patterns (catalog access, identity, etc.) based on user intent. Triggers on phrases like "create a new plugin", "scaffold a Backstage plugin", "I want to add a frontend/backend plugin", "build a new module", "new backend module", "new catalog processor", "new entity provider", "new scaffolder action", or any request to add a new package to a Backstage repo.
Use when creating or modifying LottieFiles Creator Plugins with the Creator Plugin API. Covers plugin sandbox/UI architecture, message passing, scene graph manipulation, assets, storage, animation, and plugin development workflows.
Turn a finished CodeSpring plan into the pack a paying client actually receives — a plain-English summary of what is being built, the mockup and its style guide, the plan and task links, an honest phase-by-phase build order, what was assumed, what is still open, what is needed from them, and exactly what to do next to start building. Written at a reading level a non-technical owner can follow, with no file paths and no jargon. Run it once the map, notes, PRDs and tasks exist and the plan is about to leave your hands. Triggers, "hand this over to the client", "create the client pack", "write the handover", "send them the plan", "what do I send the client", "onboard them onto this plan", "package this up for delivery".
Get an existing app into CodeSpring as a visual feature map — a small set of core features (sidebar-level) with nested sub-features and a "how it works" note each — then PRDs (via cs-build-create-prd) and Kanban tasks. Takes an OPTIONAL audit + verdict: with no audit it documents what exists; with an audit it builds the corrected target map (what the app should be) and generates fix or rebuild tasks from the findings. Detects an existing map, judges whether it is any good, and keeps or repairs it instead of duplicating it — and offers to audit first if the codebase looks broken. Triggers, "import my codebase into CodeSpring", "map my existing app", "reverse engineer my app into features", "bring this project into CodeSpring", "visualize my code in CodeSpring".
Migrate Dialogflow CX (DFCX) agents to CXAS (Customer Experience Agent Studio) agents. Use this skill when the user mentions DFCX migration, migrating agents, converting DFCX to CXAS, porting agents, agent migration, or post-migration optimization/consolidation. Four independently runnable scripts: migrate.py (1:1), stage_1.py (variable dedup + consolidation), stage_2.py (instruction state machines + tool mocks + lint + report), stage_3.py (rewires consolidated topology from source dep graph; only needed when stage_1 ran consolidation). State persists between scripts via <target>_ir.json so each can run / re-run / resume independently.
Use when running or designing an EventStorming workshop — phrases like "EventStorming", "event storming", "domain discovery workshop", "sticky notes on a wall", "big-picture EventStorming", "process modeling EventStorming", "software design EventStorming", "pivotal event", "hotspot", "should we do an EventStorming session". Defines EventStorming as Alberto Brandolini's lightweight collaborative workshop format for exploring a domain by mapping it as a sequence of past-tense events.
DDD4J Boot core development convention skill. Covers the entity inheritance chain (BaseEntity→Model<T>/PaginationEntity), DDD annotation rules (@DomainEntity is SOURCE/@DomainService≡@Service), Controller hierarchy (BaseController→BaseMapperController), Service hierarchy (IBaseService→BaseServiceImpl), return value conventions (ApiRestResponse three states success/fail/error, pagination returns rows not data), query parameter placement (beginTime/endTime/keywords/params on the Entity), internationalization NestedMessageSource, automatic audit field filling, and Dozer object mapping. Use this skill when the user writes Entity/Service/Controller, handles return value formats, or uses DDD annotations in a DDD4J project. Use together with ddd4j-jackson (serialization), ddd4j-mybatis (persistence layer), ddd4j-validation (validation), and ddd4j-satoken (authentication).
DDD4J Jackson serialization rule skill. Covers date format (yyyy-MM-dd HH:mm:ss instead of ISO-8601), null value serialization rules (arrays→[], strings→"", numbers/booleans→stay null), @Sensitive sensitive data masking (5 strategies), the dual instances of JacksonKit and WebObjectMapper (JacksonKit for Redis/ObjectMapper for Web), and ToLongDeserializer (handling frontend JS precision loss). Use when the user uses Jackson in a DDD4J project, needs to configure ObjectMapper, handles sensitive data masking, or solves frontend-backend JSON format issues. Use together with the ddd4j-core skill.
DDD4J Sa-Token authentication extension skill. Covers the StpKit extension payload pattern (write user/org/role at login, read with StpUtil.getExtra), SaTempKit one-time tokens, integration with ApiRestResponse exception handling, cross-service Same-Token configuration, and how to use @SaCheckPermission in DDD4J Controllers. Use this skill when the user uses Sa-Token for authentication in a DDD4J project, needs login extension payloads (uuid/username/orgId), or generates one-time temporary tokens. Use together with the ddd4j-core and sa-token skills.
DDD4J custom validation rule skill. Covers four custom constraints (@AllowedValues/@PhoneNumber/@NumberValue/@StringDateValue), the rule that @StringDateValue silently passes on empty strings, the constraint that validators disable default message interpolation, validation layer placement (Param/DTO rather than a standalone Validator), and libphonenumber international phone number validation. Use this skill when the user writes parameter validation, phone number/date format validation, or enum value validation in a DDD4J project. Use together with the ddd4j-core skill.
Building decay and upkeep systems for survival games. Use when implementing timer-based decay, Tool Cupboard patterns (Rust-style protection radius), resource upkeep costs, or server performance management through automatic cleanup. Balances gameplay and server health.
Use this skill whenever the user wants to call DeepSeek V4-Pro / V4-Flash (or its legacy aliases deepseek-chat / deepseek-reasoner), or you see code that imports `from openai import OpenAI` with `base_url="https://api.deepseek.com"`. This skill teaches you the 10 protocol contract rules required to avoid the 16 documented bugs (reasoning_content lifecycle 400, tool-call leakage, max_tokens runaway → V8 string limit, parallel tool delta interleave, prefix-cache invalidation, /beta endpoint remap, etc.) and ships a single drop-in `safe_init.py` script you can copy into the user's project.
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.