togetherai-skills Claude Code Plugin
Claude Code PluginClaude CodeDevelopmentAI Agents & AI App DevelopmentClaude Code Customization & WorkflowThe explanation below is AI-generated. Please verify it against the sources.
togetherai-skills is a plugin that packages 12 Agent Skills giving AI coding agents (Claude Code, Cursor, Codex, Gemini CLI) working knowledge of the Together AI platform. Each skill covers a specific Together AI product area — chat completions, images, video, audio, embeddings, fine-tuning, batch inference, evaluations, sandboxes, dedicated endpoints/containers, and GPU clusters — with SKILL.md routing guidance, reference docs, and runnable Python scripts built on the Together Python v2 SDK. According to the README, skills activate automatically when an agent detects a relevant task, requiring no explicit invocation. Installation is done via npx skills add togethercomputer/skills or by copying the together-* skill folders into an agent's skills directory. The README notes the repo targets together>=2.0.0 and points to a migration guide for users of the v1 SDK.
About the service
Together AI, per its homepage, is described as an 'AI Native Cloud' offering a full-stack platform spanning serverless and batch inference, provisioned throughput, dedicated model/container inference, GPU clusters, sandboxes, managed storage, fine-tuning, and evaluations, along with a library of open-source models.
What you can do with togetherai-skills
- Use together-chat-completions for real-time/streaming chat, tool calling, and structured outputs
- Use together-images for text-to-image generation and editing (FLUX, Kontext, LoRA)
- Use together-video for text-to-video and image-to-video generation
- Use together-audio for text-to-speech and speech-to-text (REST, streaming, WebSocket)
- Use together-embeddings for embeddings, semantic search, RAG, and reranking
- Use together-fine-tuning for LoRA, full fine-tuning, DPO, VLM, function-calling, and reasoning tuning
- Use together-batch-inference for high-volume asynchronous offline inference
- Use together-evaluations for LLM-as-a-judge model evaluation
- Use together-sandboxes for stateful remote Python execution
- Use together-dedicated-endpoints and together-dedicated-containers for single-tenant/custom Docker inference
- Use together-gpu-clusters to manage on-demand/reserved GPU clusters and storage
- Install all skills at once with npx skills add togethercomputer/skills, or copy together-* folders into Claude Code, Cursor, Codex, or Gemini CLI skill directories