Claude API
Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).
30-Second Summary
Claude API is an agent-readable workflow from anthropics/skills for claude api.
It gives the agent a trigger, ordered guidance, and source-backed checks instead of treating claude api as a generic tool.
Use it as a brief only while the linked SKILL.md remains available and the risk boundary is still accurate.
1-Minute Read
What it is
Claude API packages instructions from anthropics/skills into a reusable agent skill brief. The original source is a public SKILL.md file, so the brief can point readers back to the executable workflow instead of a product landing page.
When to use it
Use it when a user asks for help around claude api and the agent needs a repeatable workflow, checklist, or review path rather than broad background information.
How to test it first
Open the linked SKILL.md, confirm the trigger and procedure still match claude api, list the risk boundary, then ask the agent to apply the smallest read-only step to a sample task.
Watch out
Do not present this as a generic app or platform. Keep the scope tied to anthropics/skills. Risk boundary: git state, commit, branch, PR, or push actions require explicit repo-owner approval before mutation; CI/CD, deploy, release, rollback, and migration steps must start with read-only inspection and an approval gate; database, Fabric workspace, notebook, lakehouse, warehouse, or Power BI changes need credential and production-data approval; cloud resources, RBAC, credentials, tokens, and private keys must not be changed or exposed without a separate approval; Anthropic skill package; keep the scope tied to the referenced SKILL.md and bundled resources.