tokens&
For enterprises
Submit
Sign in
Build agents/Agent Skills

Agent Skills

Browse source-linked packages for your coding agent.

Create Agent SkillSubmit SKILL.mdInstall pack

The registry

129 skills from 16 vendors, each published only when its SKILL.md resolves from the vendor's repository. 8 tracked packages are held back because their source no longer resolves upstream.

JSON API
Models (39)Backend (30)Agents (28)Frontend (22)Docs (8)Payments (2)

Client coverage: Claude Code (115) · Codex (115) · Cursor (115) · Gemini CLI (103) · GitHub Copilot (85) · VS Code (85) · MCP clients (19)

Models39 skills

  • Arize dataset workflowsArizeCreate, manage, and reuse datasets for experiments, evals, and LLM regression checks.MITCodexCursorClaude Code
  • Arize evaluatorArizeDesign and run LLM-as-judge or deterministic evaluators against real application data.MITCodexCursorClaude Code
  • Arize prompt optimizationArizeUse trace data and evaluations to improve prompts and compare production behavior.MITCodexCursorClaude Code
  • Agent Platform model inferenceGoogleCall Gemini and OpenMaaS models with current SDKs, correct endpoints, authentication, and error handling.Apache-2.0CodexCursorClaude Code
  • AI Studio to Agent Platform migrationGoogleMove Gemini applications from Google AI Studio to cloud IAM, billing, telemetry, and enterprise infrastructure.Apache-2.0CodexCursorClaude Code
  • Gemini API on Agent PlatformGoogleBuild multimodal, tool-using, streaming, cached, and batch Gemini applications with current Google Gen AI SDKs.Apache-2.0CodexCursorClaude Code
  • Community model evalsHugging FaceRun inspect-ai and lighteval evaluations for model selection, regressions, and benchmarks.Apache-2.0CodexCursorClaude Code
  • Dataset Viewer workflowsHugging FaceFetch dataset metadata, split rows, filtered samples, and viewer API results for data tasks.Apache-2.0CodexCursorClaude Code
  • Hugging Face CLIHugging FacePackage model, dataset, and Space operations into a repeatable agent workflow.Apache-2.0
  • LLM trainerHugging FaceFine-tune language models with TRL, Unsloth, local hardware, or Hugging Face Jobs.Apache-2.0CodexCursorClaude Code
  • Local model runnerHugging FaceSelect GGUF models, quantization, and llama.cpp server settings for local inference.Apache-2.0CodexCursorClaude Code
  • LoRA Space builderHugging FaceBuild and publish Gradio Spaces that demonstrate user-provided LoRA adapters.Apache-2.0CodexCursorClaude Code
  • Model memory estimatorHugging FaceEstimate memory needed to load Safetensors or GGUF model weights for inference.Apache-2.0CodexCursorClaude Code
  • Sentence Transformers trainerHugging FaceFine-tune embedding models for retrieval, similarity, clustering, and classification.Apache-2.0CodexCursorClaude Code
  • Trackio experiment trackingHugging FaceLog, visualize, alert on, and retrieve ML training metrics with Trackio.Apache-2.0CodexCursorClaude Code
  • Transformers.js appsHugging FaceRun machine learning models directly in JavaScript and TypeScript browser or Node apps.Apache-2.0CodexCursorClaude Code
  • TRL training workflowsHugging FaceRun SFT, DPO, GRPO, KTO, RLOO, and reward-model training with TRL.Apache-2.0CodexCursorClaude Code
  • Vision model trainerHugging FaceTrain object detection, image classification, segmentation, and vision-language models.Apache-2.0CodexCursorClaude Code
  • Azure AI content safety for PythonMicrosoftDetect harmful text and image content with severity-aware moderation workflows in Python.MITCodexCursorClaude Code
  • Azure image analysis for PythonMicrosoftGenerate captions, tags, objects, OCR, people detection, and smart crops from images.MITCodexCursorClaude Code
  • Azure Machine Learning for PythonMicrosoftCreate and manage Azure ML workspaces, training jobs, models, datasets, compute, and pipelines.MITCodexCursorClaude Code
  • Azure multimodal extraction for PythonMicrosoftExtract semantic content from documents, images, audio, and video for RAG and automation workflows.MITCodexCursorClaude Code
  • Azure text analytics for PythonMicrosoftAdd sentiment, entity, key phrase, language, PII, and healthcare text analysis to Python apps.MITCodexCursorClaude Code
  • Azure text translation for PythonMicrosoftImplement real-time translation, transliteration, language detection, and dictionary lookup.MITCodexCursorClaude Code
  • Azure transcription for PythonMicrosoftBuild real-time and batch speech-to-text workflows with timestamps and speaker diarization.MITCodexCursorClaude Code
  • CUDA-Q onboarding guideNVIDIAInstall CUDA-Q, validate simulators and hardware targets, and build reproducible quantum applications.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuOpt installationNVIDIASelect and verify a compatible cuOpt Python, C, or REST server installation for an NVIDIA GPU environment.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuOpt numerical optimization APINVIDIASolve linear, mixed-integer, and quadratic programs with the cuOpt Python API and result diagnostics.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuOpt optimization formulationNVIDIATranslate business constraints and objectives into verifiable cuOpt mathematical programs before implementation.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuOpt routing API for PythonNVIDIABuild vehicle-routing and fleet-optimization models with constraints, objectives, and solution validation.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuPyNumeric installationNVIDIAPlan an isolated cuPyNumeric installation and verify NumPy-compatible execution and GPU usage without modifying global environments.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • cuPyNumeric migration readinessNVIDIAAudit NumPy and SciPy applications for multi-GPU cuPyNumeric migration compatibility and performance risks.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • DALI dynamic-mode pipelinesNVIDIABuild and debug GPU-accelerated data loading and preprocessing pipelines with NVIDIA DALI.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Earth2Studio installationNVIDIAChoose model extras, create an isolated environment, and verify an Earth2Studio installation for weather and climate workflows.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • NeMo Data DesignerNVIDIACreate declarative synthetic-data generation pipelines for model development, evaluation, and retrieval workflows.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • NVIDIA cuDF dataframesNVIDIAUse cuDF, pandas acceleration, Dask-cuDF, ETL, joins, and GPU dataframe patterns.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code
  • NVIDIA RAG BlueprintNVIDIADeploy, configure, troubleshoot, and manage NVIDIA RAG Blueprint services.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code
  • NVIDIA RAG evalsNVIDIARun RAGAS quality benchmarks against filesystem RAG corpora and deployed RAG endpoints.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code
  • NVIDIA RAG performanceNVIDIAProfile and load-test deployed NVIDIA RAG Blueprint servers with repeatable configs.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code

Backend30 skills

  • Cloudflare Email ServiceCloudflareSend, receive, route, and test transactional email from Workers and REST APIs.Apache-2.0CodexCursorClaude Code
  • Durable Objects builderCloudflareBuild and review stateful Workers apps with RPC, SQLite storage, alarms, and WebSockets.Apache-2.0
  • Turnstile setupCloudflareInstall Turnstile end to end with widget creation, siteverify Worker, and frontend snippets.Apache-2.0CodexCursorClaude Code
  • Workers best practicesCloudflareAuthor and review Workers code, wrangler config, bindings, secrets handling, and observability.Apache-2.0
  • Wrangler CLI guideCloudflareUse current Wrangler commands and config for Workers, KV, R2, D1, Queues, and secrets.Apache-2.0CodexCursorClaude Code
  • Elasticsearch ES|QLElasticQuery and analyze Elasticsearch logs, metrics, and business data with repeatable ES|QL workflows.Apache-2.0CodexCursorClaude Code
  • Elasticsearch onboardingElasticTurn a search use case and sample data into a working Elasticsearch search experience.Apache-2.0CodexCursorClaude Code
  • Inngest brownfield auditInngestFind fragile background work and introduce durable execution incrementally without changing existing behavior.Apache-2.0CodexCursorClaude Code
  • Inngest durable functionsInngestBuild retryable event-driven and scheduled functions that survive crashes, deploys, and serverless limits.Apache-2.0CodexCursorClaude Code
  • Inngest flow controlInngestApply concurrency, throttling, rate limits, batching, debounce, and priority to durable workflows.Apache-2.0CodexCursorClaude Code
  • Inngest middlewareInngestAdd typed logging, authentication context, dependency injection, and lifecycle behavior to functions.Apache-2.0CodexCursorClaude Code
  • Inngest project setupInngestInstall and configure Inngest in TypeScript frameworks with safe local and production boundaries.Apache-2.0CodexCursorClaude Code
  • Inngest v3 to v4 migrationInngestDetect mixed SDK usage and migrate triggers, events, realtime, invocation, and runtime settings safely.Apache-2.0CodexCursorClaude Code
  • Inngest workflow stepsInngestCompose retriable steps, sleeps, event waits, function calls, AI calls, and durable workflow state.Apache-2.0CodexCursorClaude Code
  • Azure AI Search for TypeScriptMicrosoftBuild vector, hybrid, semantic, and agentic retrieval workflows with the Azure AI Search SDK.MITCodexCursorClaude Code
  • Azure cloud solution architectMicrosoftDesign and review Azure systems against architecture patterns and Well-Architected Framework tradeoffs.MITCodexCursorClaude Code
  • Azure Cosmos DB for TypeScriptMicrosoftImplement Cosmos DB document CRUD, queries, partitioning, bulk operations, and container management.MITCodexCursorClaude Code
  • KQL query engineeringMicrosoftWrite and debug efficient Kusto queries for logs, time series, joins, anomaly detection, and analytics.MITCodexCursorClaude Code
  • cuPyNumeric parallel data loadingNVIDIAImplement scalable parallel data ingestion for multi-node, multi-GPU NumPy-compatible workloads.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan Conda installationNVIDIAInstall and verify Holoscan SDK in a compatible CUDA 13 Conda environment with platform-specific guardrails.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan container installationNVIDIASelect, pull, and verify the correct Holoscan NGC container for the host GPU and architecture.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan Debian installationNVIDIAInstall and verify native Holoscan C++ runtime and headers on supported Ubuntu and Jetson systems.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan Python installationNVIDIAInstall Holoscan Python bindings in an isolated environment and verify CUDA-matched example execution.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan setup routerNVIDIAInspect host hardware and software, select a compatible Holoscan installation path, and route to its verification workflow.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Holoscan source installationNVIDIABuild and verify Holoscan SDK from source when published packages cannot satisfy custom build requirements.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • NVIDIA Dynamo router starterNVIDIAStart or patch Dynamo router modes and run endpoint smoke checks.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code
  • Postgres best practicesSupabaseTune queries, RLS, connection pooling, and schema choices before a build slows down.MIT
  • Supabase platform guideSupabaseWork across Supabase Auth, Edge Functions, Realtime, Storage, Vectors, CLI, MCP, and RLS.MITCodexCursorClaude Code
  • Terraform best practicesTerramateReview Terraform modules, state, plans, and delivery workflows before infrastructure changes ship.MITCodexCursorClaude Code
  • Terramate best practicesTerramateUse Terramate stacks, generation, orchestration, and drift-safe infrastructure workflows.MITCodexCursorClaude Code

Agents28 skills

  • MCP builderAnthropicTurn an API or workflow into an MCP server your agent can call safely.Source review
  • Arize compliance auditArizeAudit AI agent and LLM apps for compliance gaps, tracing coverage, and remediation work.MITCodexCursorClaude Code
  • Arize instrumentationArizeAdd OpenTelemetry tracing to LLM apps with AX setup, validation, and production checks.MITCodexCursorClaude Code
  • Arize trace debuggingArizeExport and inspect traces, spans, sessions, and errors for LLM app root-cause analysis.MITCodexCursorClaude Code
  • Agents SDK build guideCloudflareCreate stateful Workers agents, MCP servers, workflows, queues, and realtime apps.Apache-2.0
  • Kibana alerting rulesElasticManage Kibana alert rules and lifecycle through REST APIs or rules-as-code workflows.Apache-2.0CodexCursorClaude Code
  • Kubernetes investigationElasticDiagnose pod, node, autoscaling, and control-plane failures from OpenTelemetry signals.Apache-2.0CodexCursorClaude Code
  • LLM observabilityElasticMonitor agent performance, token cost, response quality, and multi-step workflow behavior.Apache-2.0CodexCursorClaude Code
  • Agent Platform eval flywheelGoogleBuild eval datasets, choose quality metrics, analyze failures, and compare agent improvements systematically.Apache-2.0CodexCursorClaude Code
  • Agent Platform prompt managementGoogleCreate, retrieve, version, list, and safely delete managed prompts using repeatable Python workflows.Apache-2.0CodexCursorClaude Code
  • Agent Platform skill registryGoogleDiscover, register, version, generate, update, and inspect skills in Gemini Enterprise Agent Platform.Apache-2.0CodexCursorClaude Code
  • Gemini Interactions APIGoogleCreate stateful multi-turn, background, streaming, structured-output, and tool-calling agent workflows.Apache-2.0CodexCursorClaude Code
  • Gemini managed agents APIGoogleCreate and manage stateful Gemini agents with files, skills, tools, and server-managed conversations.Apache-2.0CodexCursorClaude Code
  • Google Agents CLI onboardingGoogleInstall and use the Agents CLI across local development, testing, evaluation, deployment, and monitoring.Apache-2.0CodexCursorClaude Code
  • Hub tool builderHugging FaceCreate scripts and tools that chain Hugging Face Hub API data safely.Apache-2.0CodexCursorClaude Code
  • Hugging Face MCPHugging FaceUse Hugging Face Hub MCP tools for models, datasets, Spaces, papers, and compute jobs.Apache-2.0MCP clientsCodexCursor
  • Durable Inngest agentsInngestBuild crash-safe AI agents, tool workflows, approvals, realtime progress, and provider-aware execution.Apache-2.0CodexCursorClaude Code
  • Inngest Agent EvalsInngestScore durable agent outcomes, compare experiments, and debug production workflows with traceable eval signals.Apache-2.0CodexCursorClaude Code
  • Azure AI Projects for TypeScriptMicrosoftBuild Foundry applications with project clients, agents, deployments, datasets, indexes, and evaluations.MITCodexCursorClaude Code
  • Azure hosted agents for PythonMicrosoftBuild persistent Azure AI Foundry agents with hosted tools, MCP integrations, streaming, and structured outputs.MITCodexCursorClaude Code
  • Azure Monitor OpenTelemetry for TypeScriptMicrosoftAdd distributed traces, metrics, and logs to Node.js services with Azure Monitor and OpenTelemetry.MITCodexCursorClaude Code
  • Azure real-time voice AI for PythonMicrosoftBuild bidirectional voice agents with WebSockets, VAD, tools, avatars, and live transcription.MITCodexCursorClaude Code
  • GitHub Copilot SDK builderMicrosoftBuild programmatic Copilot integrations with sessions, tools, streaming, hooks, MCP, and BYOK providers.MITCodexCursorClaude Code
  • AI-Q Blueprint deploymentNVIDIAInstall, run, validate, troubleshoot, and stop a local or self-hosted NVIDIA AI-Q Blueprint environment.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • DeepStream pipeline generatorNVIDIAGenerate inspectable DeepStream video-analytics pipelines from application, model, and throughput requirements.Apache-2.0 / CC-BY-4.0CodexCursorClaude Code
  • Agent Browser automationVercelAutomate browser navigation, form filling, screenshots, scraping, and web app verification.Apache-2.0CodexCursorClaude Code
  • AI SDK builderVercelBuild AI-powered features with current AI SDK APIs, docs, examples, and package inspection.Source reviewCodexCursorClaude Code
  • Vercel cost and performance optimizerVercelAudit deployed Vercel projects for metric-backed cost, performance, caching, and reliability wins.Source review

Frontend22 skills

  • React Native library creatorCallstackCreate a React Native library with the Callstack incubator workflow and package checks.MITCodexCursorClaude Code
  • React Native upgraderCallstackUpgrade React Native projects with version checks, migration steps, and validation commands.MITCodexCursorClaude Code
  • Web performance auditCloudflareMeasure Core Web Vitals, identify resource bottlenecks, and prioritize page-speed fixes.Apache-2.0
  • Kibana dashboardsElasticCreate, version, and deploy Kibana dashboards and visualizations as repeatable artifacts.Apache-2.0CodexCursorClaude Code
  • Gradio app builderHugging FaceBuild and edit Gradio apps, demos, chatbots, layouts, and event listeners.Apache-2.0CodexCursorClaude Code
  • Hugging Face SpacesHugging FaceBuild, deploy, debug, and maintain Gradio, Docker, Static, ZeroGPU, and hardware-backed Spaces.Apache-2.0CodexCursorClaude Code
  • ZeroGPU SpacesHugging FaceConfigure Gradio Spaces with ZeroGPU decorators, requirements, and runtime checks.Apache-2.0CodexCursorClaude Code
  • Inngest realtime workflowsInngestStream typed durable workflow and agent progress to React interfaces with retry-safe publishing.Apache-2.0CodexCursorClaude Code
  • Application Insights web instrumentationMicrosoftInstrument browser apps for RUM, dependencies, exceptions, custom events, and correlated agent traces.MITCodexCursorClaude Code
  • Frontend design reviewMicrosoftReview UI code for accessibility, responsive behavior, design-system compliance, craft, and trust.MITCodexCursorClaude Code
  • Qt C++ code reviewThe Qt CompanyReview Qt 6 C++ ownership, threading, model contracts, APIs, errors, and performance.BSD-3-ClauseCodexCursorClaude Code
  • Qt CMake project builderThe Qt CompanyGenerate and update Qt 6 CMake projects, targets, sources, resources, and QML modules.BSD-3-ClauseCodexCursorClaude Code
  • Qt QML code reviewThe Qt CompanyAudit Qt 6 QML bindings, layouts, loaders, delegates, states, and performance.BSD-3-ClauseCodexCursorClaude Code
  • Qt QML codingThe Qt CompanyWrite, fix, refactor, and optimize Qt 6 QML with focused framework guardrails.BSD-3-ClauseCodexCursorClaude Code
  • Qt QML profilerThe Qt CompanyProfile Qt Quick frame time, bindings, signals, rendering, memory, and pixmap caches.BSD-3-ClauseCodexCursorClaude Code
  • Qt QML test runnerThe Qt CompanyBuild and run Qt Quick tests through qmltestrunner or CTest and report failures.BSD-3-ClauseCodexCursorClaude Code
  • Qt Quick test generatorThe Qt CompanyGenerate Qt Quick Test coverage with TestCase, SignalSpy, and asynchronous assertions.BSD-3-ClauseCodexCursorClaude Code
  • Qt UI designThe Qt CompanyDesign and audit Qt interfaces for desktop, embedded MPU, and MCU targets.BSD-3-ClauseCodexCursorClaude Code
  • Deploy to VercelVercelDeploy projects as preview deployments with CLI setup, project linking, and production safeguards.Source review
  • React performance best practicesVercelReview React and Next.js code for performance, server/client boundaries, bundles, and data fetching.MIT
  • React view transitionsVercelImplement View Transition API animations, shared element transitions, and reduced-motion fallbacks.Source review
  • Web design guidelines reviewVercelReview UI code for accessibility, UX, and interface guideline compliance.Source review

Docs8 skills

  • Agent Platform RAG engineGoogleManage RAG corpora and files, retrieve grounded context, and generate corpus-grounded responses.Apache-2.0CodexCursorClaude Code
  • Hugging Face PapersHugging FaceRead paper metadata, linked models, datasets, Spaces, and project pages through the Papers API.Apache-2.0CodexCursorClaude Code
  • Azure document translation for PythonMicrosoftBatch-translate Word, PDF, Excel, and PowerPoint files while preserving document formatting.MITCodexCursorClaude Code
  • Microsoft docs retrievalMicrosoftGround Microsoft platform answers in official documentation through Learn MCP or its public CLI fallback.MITCodexCursorClaude Code
  • NVIDIA NeMo RetrieverNVIDIAIndex and query multi-file document corpora with the NeMo Retriever CLI.Apache-2.0 AND CC-BY-4.0CodexCursorClaude Code
  • OpenAI docs guideOpenAIUse official OpenAI docs as the retrieval source when building with OpenAI APIs.Source review
  • Qt C++ documentationThe Qt CompanyGenerate Markdown API references for Qt C++ classes, modules, utilities, and entry points.BSD-3-ClauseCodexCursorClaude Code
  • Qt QML documentationThe Qt CompanyGenerate standalone Markdown reference documentation for QML components and applications.BSD-3-ClauseCodexCursorClaude Code

Payments2 skills

  • Stripe API upgraderStripeUpgrade Stripe API versions and SDKs with version-specific migration guardrails.MITCodexCursorClaude Code
  • Stripe integration guideStripeBuild payment flows, API integrations, webhooks, and billing behavior with agent-ready guardrails.MIT
tokens&

Build better AI stacks, claim useful opportunities, and give AI infrastructure companies a source-labeled adoption readout they can trust.

For buildersFor enterprises

Product

  • For builders
  • Category rankings
  • Startup credits and perks
  • Agent Skills
  • Platform
  • Submit project, tool, product, or perk

Enterprise

  • Start free company workspace

Community

  • Community
  • Newsletter
  • Events
Xin

© 2026 tokensand, LLC. All rights reserved.

  • Terms
  • Privacy
  • Security
  • Data Processing
  • Status