What it is
A skill that gives an assistant reference material for building LLM applications with the Claude API and Anthropic SDKs. It detects the project language, picks a surface (single API call, tool use, Managed Agents, or Agent SDK), and sets defaults such as the model string, adaptive thinking and streaming. It also exposes subcommands (migrate, prompt-audit, upgrade, cost-optimize, build-eval, preserved-thinking-migration, hillclimb) that each point to a dedicated guide.
Who it's for
- Developers writing or modifying code that calls Claude through the Anthropic SDK
- Teams migrating existing Claude API code to a newer model
- Developers choosing between the Claude API, Tool Runner, Managed Agents, and the Claude Agent SDK
- Teams who want to audit prompts, reduce Claude API costs, or build and hill-climb evals
Requirements
Requirements
- A project in a supported SDK language (Python, TypeScript/JavaScript, Java/Kotlin/Scala, Go, Ruby, C#, PHP), or cURL/raw HTTP when no official SDK exists
- Claude/Anthropic as the target provider; the skill stops if non-Anthropic provider markers are found
- Network access to WebFetch SDK repositories listed in
shared/live-sources.mdwhen a binding is not documented locally - An Admin API key is optionally used by
cost-optimizeto measure token usage and cost
Examples
Default model and thinking settings
textmodel: claude-opus-5-5
thinking: {type: "adaptive"}What it does: The skill's stated defaults: model string claude-opus-5-5 and adaptive thinking for anything remotely complicated. It also defaults to streaming for long input or output, or high max_tokens.
Eager input streaming for client tools
texteager_input_streaming: trueWhat it does: Set on each user-defined (client) tool in a streaming request so large tool inputs stream as generated. The client must validate parsed tool inputs against the schema before running them.
Migrate to a newer model
Prompt/claude-api migrateExpected output: Runs the migrate subcommand, which reads shared/model-migration.md, confirms scope, classifies files, then applies per-target changes and audits prompts.
Upgrade the Python SDK
Prompt/claude-api upgrade python sdk src/Expected output: Runs the upgrade subcommand with a language and scope, following python/claude-api/sdk-upgrade.md for the anthropic 0.x to 1.x upgrade.
Audit prompts for dated patterns
Prompt/claude-api prompt-auditExpected output: Runs a non-interactive audit of prompts, tool descriptions, skills, and agent config files, producing a findings report and a proposed diff.
Pros & cons
Pros
- Pro:Warns about API drift (e.g. thinking config, web search tool types, Files API/Skills namespaces) and treats the skill's language files as authoritative
- Pro:Requires official SDK usage and forbids guessing SDK signatures, falling back to documented patterns or live SDK repos
- Pro:Includes subcommands for migration, prompt audit, SDK upgrade, cost optimization, eval building, and hill-climbing
- Pro:Helps choose between a single call, tool runner, manual loop, Managed Agents, and the Claude Agent SDK
Cons
- Con:Refuses to edit files with non-Anthropic provider markers and asks the user first, so it is Claude-only
- Con:The SDK
upgradeguide currently covers only the Python SDK (0.x to 1.x); other languages are pointed to their CHANGELOGs - Con:Several covered features (Tool Runner, Managed Agents) are beta
Images
Related
Skill 
Skill algorithmic-art
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
Skill brand-guidelines
Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
Skill canvas-design
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.
Skill 
Skill doc-coauthoring
Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.