Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Claude API Reference Skill

Official

Official Anthropic skill that grounds Claude Code in current API facts — model IDs, pricing, streaming, tool use, agents, caching and migrations.

Dev & CodingIntermediate★ 167,132⑂ 19,920AI score 9/10Last updated: Aug 7, 2026

What it does

A reference skill that forces Claude Code to look up the Anthropic API instead of answering from memory before writing or editing LLM code.

  • Current model IDs with input/output pricing per 1M tokens and context windows
  • Up-to-date parameter rules: thinking: {type: "adaptive"}, output_config.effort, plus which shapes now return 400 on which models
  • Automatic language detection (Python, TypeScript, Java, Go, Ruby, C#, PHP, cURL) so only the relevant {lang}/ docs are read
  • A decision table for single call vs. workflow vs. custom agent (manual loop or Tool Runner) vs. Managed Agents, including the harness-vs-deployment mental model
  • Guidance on prompt caching, token counting, structured outputs, MCP, server-side tools, and model migration via /claude-api migrate
  • Guardrail: if the file or project targets another provider (OpenAI, Gemini, Mistral…), the skill stops and asks instead of injecting Anthropic SDK calls

Who it's for

  • Backend/AI engineers shipping production features on the Anthropic SDK
  • Anyone tired of generated code using stale patterns like budget_tokens or date-suffixed model IDs
  • Tech leads comparing cost, context limits and agent architectures
  • Teams migrating older Claude API code to newer models

Example uses

  1. "Add a document-summarization endpoint to this FastAPI service using Claude" → detects Python, writes anthropic SDK code with streaming and adaptive thinking.
  2. "For a nightly agent, should I use the Tool Runner or Managed Agents?" → compares the four agent approaches and points to scheduled Managed Agents deployments.
  3. "/claude-api migrate" → confirms file scope and target model first, then applies the breaking-change checklist from the migration guide step by step.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/claude-api folder from the GitHub repo anthropics/skills into my ~/.claude/skills/claude-api/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/anthropics/skills.git /tmp/anthropic-skills && mkdir -p ~/.claude/skills && cp -r /tmp/anthropic-skills/skills/claude-api ~/.claude/skills/claude-api

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal (Terminal on macOS; WSL recommended on Windows).
  2. Clone the official repo into a temp folder: git clone https://github.com/anthropics/skills.git /tmp/anthropic-skills
  3. Create your skills directory: mkdir -p ~/.claude/skills
  4. Copy just this skill: cp -r /tmp/anthropic-skills/skills/claude-api ~/.claude/skills/claude-api
    • Make sure the subfolders (python/, typescript/, shared/, etc.) come along with SKILL.md — the skill depends on them.
  5. Run claude inside your project and confirm it loads with /skills or by invoking /claude-api migrate.
  6. Set up credentials: export ANTHROPIC_API_KEY, or run ant auth login (the skill checks ant auth status first).
  7. Re-run git pull in the cloned repo and repeat step 4 periodically so the cached model and pricing tables stay current.
View source on GitHub ↗