anthropic-api-core

anthropic-api-core

Claude API (platform.claude.com) — Official reference covering text generation, streaming, thinking, prompt caching, structured outputs, Message Batches, vision/PDF, and the Files API, centered on the Messages API. Includes error codes, rate limits, beta headers, and how to access via Amazon Bedrock / Google Cloud Vertex AI / Microsoft Foundry / Claude Platform on AWS.

Tool use・Agent Skills (SKILL.md)・MCP connector は anthropic-api-tools-mcp, for Claude Code itself refer to anthropic-claude-code、 Agent SDK は anthropic-agent-sdk (this skill handles how to call the Claude API itself).

Directory structure

skills/anthropic-api-core/
  SKILL.md
  references/
    getting-started/
      README.md
      get-started.md
      get-api-key.md
      intro.md
    api-basics/
      README.md
      beta-headers.md
      beta.md
      claude-platform-on-aws-iam-actions.md
      errors.md
      ip-addresses.md
      overview.md
      rate-limits.md
      service-tiers.md
      supported-regions.md
      versioning.md
    messages/
      README.md
      batch-processing.md
      citations.md
      embeddings.md
      fallback-credit.md
      fast-mode.md
      files.md
      handling-stop-reasons.md
      mid-conversation-system-messages.md
      multilingual-support.md
      overview.md
      pdf-support.md
      refusals-and-fallback.md
      search-results.md
      streaming.md
      structured-outputs.md
      task-budgets.md
      token-counting.md
      vision-coordinates.md
      vision.md
      working-with-messages.md
    thinking/
      README.md
      effort.md
      extended-thinking.md
      mid-conversation-effort-example.md
      thinking-steering-and-cost.md
      thinking-tool-workflows.md
      thinking-troubleshooting.md
      thinking.md
    caching-context/
      README.md
      cache-diagnostics.md
      compaction.md
      context-editing.md
      context-windows.md
      prompt-caching.md
    endpoints/
      README.md
      batches-cancel.md
      batches-create.md
      batches-delete.md
      batches-list.md
      batches-results.md
      batches-retrieve.md
      completions-create.md
      files-delete.md
      files-download.md
      files-list.md
      files-retrieve_metadata.md
      files-upload.md
      messages-count_tokens.md
      messages-create.md
      models-list.md
      models-retrieve.md
    platforms/
      README.md
      claude-in-amazon-bedrock.md
      claude-in-microsoft-foundry.md
      claude-on-amazon-bedrock-legacy.md
      claude-on-vertex-ai.md
      claude-platform-on-aws.md
  samples/
    README.md
    basic-message.md
    streaming.md
    extended-thinking.md
    batch-processing.md
    prompt-caching.md
    structured-outputs.md
    vision-pdf.md
    token-counting.md
  scripts/
    README.md
    setup.md
    curl-endpoints.md
    headers.md

Exploration procedure

Derive the category from the task, then identify the target page in the category's README.md:

1. Find the category corresponding to the task in the mapping table below
2. of that category references/{category}/README.md（samples/ scripts/ (the README.md directly under it) to identify the target page
3. Read the .md of the relevant page to check the details

Task → Category Mapping

Task	Category	Reference README
Want to know how to obtain an API key, make the first API call, and choose between the Messages API and Managed Agents	getting-started	references/getting-started/README.md
Want to know about error codes, rate limits, beta headers (anthropic-beta) · API version (anthropic-version), service tiers, and supported regions	api-basics	references/api-basics/README.md
Want to generate text with the Messages API, handle streaming (SSE), vision/PDF, citations, search results, embeddings, and stop reasons	messages	references/messages/README.md
Want to know about structured outputs (output_config.format), Message Batches, the Files API, fast mode, task budgets, and multilingual support	messages	references/messages/README.md
Want to know about thinking (adaptive thinking / effort), legacy extended thinking (fixed budget_tokens), tool integration with thinking, and troubleshooting	thinking	references/thinking/README.md
prompt caching（cache_control), context editing, compaction, context windows, and cache diagnostics	caching-context	references/caching-context/README.md
Want to check the request/response specifications (curl equivalent) for the Messages / Batches / Files / Models endpoints	endpoints	references/endpoints/README.md
Want to call the Claude API via Amazon Bedrock / Google Cloud Vertex AI / Microsoft Foundry / Claude Platform on AWS	platforms	references/platforms/README.md
Want to know typical usage (basic calls, streaming, thinking, batches, prompt caching, structured outputs, vision/PDF, token counting)	samples	samples/README.md
Want to know about API key setup, SDK/CLI installation, calling endpoints with curl, and header specification examples	scripts	scripts/README.md