Skills
99,064 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Use this skill to ensure Jetpack Compose performance numbers reflect production reality by measuring against a release variant with R8 enabled, Live Literals disabled, and Compose Compiler reports read from the release output directory. Covers why debug builds lie (interpreted Compose runtime, JIT warmup, Live Literals constant-getters), how to set up a release-with-symbols measurement build, and how to wire Macrobenchmark, Compose Compiler reports, Layout Inspector, simpleperf, and Android Studio Profiler against it. Cited result is roughly 75 percent startup gain and 60 percent frame-render gain debug to release. Use when the developer reports "slow startup", "jank", "dropped frames", "high recomposition count", or quotes timings from `assembleDebug`, Layout Inspector, or `CompilationMode.None`. Use when reviewing a perf bug, setting up a CI perf gate, or before filing a perf regression.
Procedural fields for coherent scalar and vector causes in Three.js WebGPU/TSL. Use when a task needs shared causes across materials, geometry, or compute; must choose direct evaluation versus a baked field; needs deterministic CPU/TSL parity and footprint filtering; or needs metric signed-distance terrain and coast analysis.
Couple Three.js WebGPU/TSL rain, snow, and wet surfaces. Use for airborne precipitation motion, physical surface deposition, persistent snow or wetness state, weather-driven ripple normals, or impact splashes.
Synthesize broad-band offshore seas with spectral FFT cascades in Three.js WebGPU/TSL. Use for homogeneous directional wind sea or swell, transported foam history, bounded CPU surface queries, or offshore forcing of a separate coastal model.
Build multi-party calls using Twilio Conference. Covers warm transfer, cold transfer, coaching (whisper), hold vs mute, participant modes, and supervisor barge. Use this skill for any contact center, support line, or scenario requiring transfers, holds, or multi-party calls.
Planning skill for building customer service and support systems. Qualifies the developer's needs across the support ladder (self-service → AI agents → contact center), channel mix, and scale to recommend the right Twilio architecture. Handles both "build me a call center" and "add an IVR to my existing support line."
Implement or modify TypeScript SDK behavior in @composio/core or shared TypeScript packages, including tools, toolkits, sessions, auth configs, connected accounts, modifiers, and generated SDK surfaces. Use for TS runtime/API work; pair with typescript-testing for verification and cross-sdk-parity when Python must match.
Use this skill to choose the cheapest correct effect API in Jetpack Compose — `LaunchedEffect`, `DisposableEffect`, `SideEffect`, `rememberUpdatedState`, and skydoves/compose-effects' `RememberedEffect` and `ViewModelStoreScope`. Covers stale-callback bugs in long-lived `LaunchedEffect`, setup/teardown for non-coroutine subscribers, avoiding a coroutine scope just to react to a key change, and per-row ViewModels in a `LazyColumn`. Trigger when the user mentions LaunchedEffect, DisposableEffect, RememberedEffect, SideEffect, rememberUpdatedState, ViewModelStoreScope, effect restarts unexpectedly, leaked listener, or per-item ViewModel.
High-level map of the Venice AI API - base URL, which auth mode each endpoint accepts (API key, x402 wallet, or none), endpoint categories (including decisions, voice changer, and retired routes), response headers (rate limit, balance, deprecation, x402), pricing model, error shape, and versioning. Load this first when starting any Venice integration.
Authenticate to the Venice API with a Bearer API key or with an x402 / SIWX wallet (EVM on Base, including EIP-1271 smart wallets, or Ed25519 on Solana). Covers which endpoints accept which scheme, the SIGN-IN-WITH-X header format, the SIWE and Solana message fields, the enforced TTL / clock-skew / nonce rules, every X402_SIGN_IN_* error code, the venice-x402-client SDK, and how to choose between the two modes
Use when work needs isolation from the current workspace — when /execute (mode=solo) or the /dispatch funnel must run a phase on its own branch without touching the main working tree. Detects existing isolation, creates a fork via `doctrine worktree fork` (the sole creation+provision+bind verb; coordination tier excluded), runs the commit-before-spawn / branch-point / baseline guards, and hands back the fork branch.
Use when managing an AI Hub account, API keys, balance, usage, or API endpoints. Use when the user says “AI Hub,” “add AI credits,” “create API key,” “check AI usage,” “auto-recharge,” “AI Hub endpoint,” “AI Hub base URL,” “how to use AI Hub API,” “LLM API,” “AI API,” “OpenAI compatible,” “Anthropic API,” “GPT,” “Claude,” “Gemini,” “DeepSeek,” or “Grok” in the context of Zeabur.
Use when you need to mine a conversation, session transcript, or design discussion for architectural decisions before writing ADRs. Identifies problem-solution pairs, trade-off debates, technology choices, and explicit "[ADR]" tags. Triggers on "what decisions did we make", "extract decisions from this chat", "find the choices in our discussion", or "summarize architectural decisions". Also useful after long planning sessions to capture decisions that were made implicitly. Does NOT write ADR documents — use adr-writing or write-adr for that.
Designs agent-native applications where agents are first-class citizens with full tool parity, atomic primitives, and explicit completion signals. Covers tool design, context injection, agent-to-UI communication, and mobile checkpoint/resume patterns. Use when architecting an agentic system, designing tool surfaces, building agent-aware UI, implementing context.md patterns, or asking “how do I make my app agent-native.”
Generate architecture diagrams as fully editable SVG with native AWS, Azure, and GCP icons for cloud diagrams, or hand-drawn generic icons for everything else. Optionally deliver a self-contained editable draw.io mxGraph companion. Deterministic layout computes zone nesting and orthogonal routing instead of hand-placed coordinates. Triggers on: "architecture diagram", "infra diagram", "system diagram", "deployment diagram", "topology diagram", "draw architecture", "AWS diagram", "Azure diagram", "GCP diagram", "cloud infrastructure diagram", "VPC diagram", "draw my AWS setup". Use when a user wants a static architecture diagram they can still edit afterward in Figma, Illustrator, Inkscape, or draw.io. NOT for architecture reviews, use architecture-reviewer.
Use this skill when a Compose tree uses SubcomposeLayout, BoxWithConstraints, or Scaffold and the developer reports extra measure passes, slow first frame, or layout passes running content composition repeatedly. Covers why SubcomposeLayout composes its slots during the measure phase, why BoxWithConstraints forces a subcomposition for every new Constraints value, why nesting Scaffold or BoxWithConstraints multiplies the cost, when a custom Layout or Modifier.layout { } replaces SubcomposeLayout, and how to use SubcomposeLayoutState's slot reuse policy and precompose APIs when SubcomposeLayout is genuinely required. Use when the developer mentions BoxWithConstraints, SubcomposeLayout, Scaffold, "extra measure pass", "double measurement", "first frame slow", "subcompose", or notices that wrapping content in BoxWithConstraints regresses scroll perf inside a LazyColumn.
Use this skill when writing or fixing an expression in a BPMN process, DMN decision, or Camunda form: gateway conditions, input and output mappings, timer durations and cycles, date math, list operations, null checks, or FEEL errors. Writes and evaluates FEEL for Camunda 8. Not for application code in JavaScript, Java, Python, or other languages
Guide Catlass operator performance tuning. Workflow: read the catlass optimization guide, obtain/update the profiler baseline, modify tiling according to the guide, rebuild, **be sure to generate and display a performance comparison report**, and iterate on the comparison. Follow the catlass documentation for tuning strategies. Ask follow-up questions if requirements are unclear.
One-shot efficiency audit of a file, diff, or whole repo across BOTH axes at once: over-engineered code (reinvented stdlib, needless abstractions, speculative config) AND bloated prose (verbose comments, padded docstrings, redundant doc sections). Neither a pure code-minimizer nor a pure prose compressor does both in one pass. That's the point. Ranked report, biggest saving first; changes nothing. Use when the user says "chisle audit", "/chisle-audit", "audit this for bloat", "what can I cut", "review this PR for over-engineering and verbosity".
Systematic code review patterns covering security, performance, maintainability, correctness, and testing — with severity levels, structured feedback guidance, review process, and anti-patterns to avoid. Use when reviewing PRs, establishing review standards, or improving review quality.
Document solved problems for team reuse. Provides process knowledge for /ia-compound. Use when documenting a resolved issue, writing up lessons learned, capturing a post-mortem, adding to the knowledge base, or building searchable institutional knowledge after debugging.
Event Storming workshop facilitation guide — collaborative domain exploration methodology with 6-step process (chaos exploration/timeline/pivotal events/commands & actors/aggregate discovery/bounded context), standardized sticky note color conventions (orange/blue/yellow/green/pink/purple), workshop preparation and facilitation tips, and digital tool recommendations. Use when user asks about event storming, 事件风暴, workshop, 工作坊, domain exploration, collaborative modeling, or needs to organize a DDD discovery workshop.
Use G2 hardware features in Even Hub apps — microphone audio capture (glasses or phone mic), IMU motion data, location (one-shot and continuous), phone album picker / camera capture, device info, user info, and local storage. Use when working with audio, IMU, location, photos, battery, wearing detection, or persistent storage.
Use this skill when the user is doing hands-on DOCA Comch work on a host + BlueField pair — bringing up host ↔ DPU PCIe control-plane messaging, picking server (DPU) vs client (host) roles, choosing slow-path send-task / recv-callback vs fast-path producer / consumer, querying max-msg-size or max-clients capabilities, registering connection callbacks, or debugging DOCA_ERROR_* returns from the Comch API. Trigger even when the user does not explicitly mention "DOCA Comch" or "Comm Channel" (renamed in DOCA 2.5) — typical implicit phrasings include "send a control message from host to BlueField over PCIe", "DPU can't see the host representor", "DOCA_ERROR_NOT_PERMITTED on server_create", "DOCA_ERROR_AGAIN on task_send submit", "connect callback never fires", or "stream bulk data from a host driver to a DPU agent". Refuse and route elsewhere for installing DOCA itself, BFB / firmware bring-up, non-Comch DOCA libraries, or deploying Comch apps at scale — those belong to other skills.
This skill guides development of full-stack features on EdgeOne Pages — Edge Functions, Cloud Functions (Node.js / Go / Python runtimes), Middleware, KV Storage, and local dev workflows. It should be used when the user wants to create APIs, serverless functions, middleware, WebSocket endpoints, or full-stack features specifically on EdgeOne Pages — e.g. "create an API", "add a serverless function", "write middleware", "build a full-stack app", "add WebSocket support", "set up edge functions", "use KV storage", "create a Go API", "build a Python backend", "use Flask/FastAPI/Gin on EdgeOne Pages". Do NOT trigger for framework-native features (Next.js API routes, Next.js middleware, Nuxt server routes) or generic Express/Koa development outside an EdgeOne Pages project. Do NOT trigger for deployment — use edgeone-pages-deploy instead. Do NOT trigger for other platforms (Cloudflare Workers, Vercel Functions, AWS Lambda).
Orchestrates Gang of Four design pattern implementations across 8 languages. Analyzes user problems to select optimal pattern(s) from creational, structural, and behavioral categories. Provides concrete, production-ready implementations with explanations and trade-offs.
When the user wants to build GTM automation with code, design workflow architectures, use AI agents for GTM tasks, or implement the 'architecture over tools' principle. Also use when the user mentions 'GTM engineering,' 'GTM automation,' 'n8n,' 'Make,' 'Zapier,' 'workflow automation,' 'Clay API,' 'instruction stacks,' 'AI agents for GTM,' or 'revenue automation.' This skill covers technical GTM infrastructure from workflow design through agent orchestration.
Set up and improve harness engineering (AGENTS.md, docs/, lint rules, eval systems, project-level prompt engineering) for AI-agent-friendly codebases. Triggers on: new/empty project setup for AI agents, AGENTS.md or CLAUDE.md creation, harness engineering questions, making agents work better on a codebase. ALSO triggers when users are frustrated or complaining about agent quality — e.g. 'the agent keeps ignoring conventions', 'it never follows instructions', 'why does it keep doing X', 'the agent is broken' — because poor agent output almost always signals harness gaps, not model problems. Covers: context engineering, architectural constraints, multi-agent coordination, evaluation, long-running agent harness, agent loops (autonomous operation, verifiers, stop conditions), agent memory (cross-session memory, consolidation), and diagnosis of agent quality issues
Sui Kiosk and NFT trading primitives. Use when writing, reviewing, or debugging Move code that uses Kiosk for commerce, NFT trading, asset listing, TransferPolicy enforcement, or building Kiosk Apps and extensions. Also use when the user asks about asset states (PLACED, LOCKED, LISTED), borrowing patterns, purchase flows, or the kiosk_extension permissioned apps API.
Check and update tool versions by scanning npm/brew/pip/uv/GitHub/plugins/skills, comparing installed and latest versions item by item, and generating a numbered report. Updates are performed only after confirmation. Use when the user says “检查更新”, “工具版本”, “看看哪些该更新了”, “brew有没有过期的”, “opencli是不是该更新了”, or mentions tool versions or environment updates.
Migrate to Langfuse from another LLM observability/evals platform (LangSmith, Arize AX, Phoenix, Braintrust, Helicone, Promptfoo, ...). A one-off vendor switch: live tracing cutover plus transfer of historical traces, datasets, prompts, and evaluators. Not for anything Langfuse-to-Langfuse — SDK or version upgrades, project/region moves, self-hosted to Cloud — and not for moving in-code prompts into Langfuse; the main langfuse skill covers those.
(NS) Ad-hoc coding — bug fixes, refactors, scripts, migrations — without full SDD. Entry priority 5: \"just implement this\", \"quick fix\", or coding without execution-handoff (C2 under ns-autonomous). Do NOT: ISSUE_URL (ns-execution-gitlab-issue), multi-day/version (ns-spec-driven), product bug report / screen expected-vs-actual **without** diagnosis_complete (ns-bug-report-diagnosis), diagnosis-only stack/CI (ns-investigator), when execution-handoff.md exists. Do NOT generate requirements/tasks/handoff.
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.