Give your AI IDE the agent playbook included with Premium
Install the AISDK Agents skill so Cursor, Claude Code, and other skill-aware assistants stop guessing at architecture and start from proven AI SDK patterns: tool loops, approval flows, artifacts, streaming UI, evaluator-optimizer systems, and provider setup.
One command. A better starting point for every agent build.
Premium customers get a short-lived install command. Run it once, and your assistant can reference the architecture notes included with AI SDK Agents Premium.
$ npx skills add https://aisdkagents.com/skill/<token> --skill aisdkagentsInstall the same skill wherever your agent work happens
The AISDK Agents skill is packaged for skill-aware coding agents and terminal assistants, including Amp, Antigravity, Codex, Cursor, Droid, GitHub Copilot, KiloCode, OpenCode, VS Code, Warp, Windsurf, and Claude Code.
Amp
Skill install target
Antigravity
Skill install target
Codex
Skill install target
Cursor
Skill install target
Droid
Skill install target
GitHub Copilot
Skill install target
KiloCode
Skill install target
OpenCode
Skill install target
VS Code
Skill install target
Warp
Skill install target
Windsurf
Skill install target
Claude Code
Skill install target
The full AI SDK agent reference package
The skill bundles the decision tree, current AI SDK grounding, anti-pattern guidance, implementation skeletons, focused pattern deep dives, and broad reference maps included with Premium.
Pattern decision tree
Pattern reference
Chooses between routing, chains, orchestration, sub-agents, human approval, artifacts, structured output, and advanced tools.
Current AI SDK APIs
Pattern reference
Grounds answers in current API naming and when to use streamText, generateObject, streamObject, tool, and ToolLoopAgent.
Agent architecture
Pattern reference
Covers routing, orchestration, sub-agents, sequential chains, parallel processing, workers, and domain agents.
Tools and repair
Pattern reference
Designs typed tools, multi-step loops, lifecycle hooks, preliminary results, context passing, dynamic tools, and repair flows.
Human in the loop
Pattern reference
Adds confirmation, approval, multiple-choice inquiry, text inquiry, and resume flows where users need control.
Artifacts
Pattern reference
Builds chat-plus-artifact systems for text, code, tables, charts, maps, slides, diagrams, canvases, and generated files.
Streaming UI
Pattern reference
Uses data parts, custom UI messages, transport setup, status states, typed tool UI, and responsive chat surfaces.
Evaluator optimizer
Pattern reference
Let agents critique and revise outputs before they reach the user.
Structured output
Pattern reference
Constrain final answers with Output.choice, Output.array, object schemas, queues, forms, JSON specs, and deterministic data extraction.
Plan and implement
Pattern reference
Splits work into planning and implementation phases with plan artifacts, todo state, requirements gathering, and execution loops.
Providers and env
Pattern reference
Map providers, environment variables, setup complexity, and required services.
Current skeletons
Pattern reference
Includes copy-paste starting points for useChat, ToolLoopAgent routes, tool approval, human inquiry, dynamic tools, artifacts, and structured output.
Anti-pattern checks
Pattern reference
Flags overbuilt orchestration, stale API choices, unnecessary sub-agents, unsafe approval flows, artifact misuse, and risky filesystem tools.
Source evidence
Pattern reference
Points the assistant toward proven pattern references without installing source code.
From Premium access to a skill-aware assistant
Upgrade to Premium
The skill is included with Premium, alongside the AI pattern library.
Copy a short-lived command
The site signs a temporary token and turns it into a Skills CLI command.
Run it in your terminal
The CLI verifies the token, downloads the skill, and adds the playbook to your assistant.
Stop asking your assistant to rediscover the pattern
Upgrade to Premium to install the AISDK Agents skill, then use it whenever you build agents, chat apps, tool-calling workflows, artifact systems, or streaming AI interfaces.