Skip to content
HUIAHUB

anthropics/skills/skills/claude-api

Claude API

Reference for Claude API and Anthropic SDK work, including models, pricing, parameters, streaming, tools, MCP, agents, caching, token counting and migrations.

Agent skillAgent skills
View the skill on GitHub (opens in a new tab)

What Claude API does

The skill first checks whether the task belongs to Claude or Anthropic. It scans the target file or project for another provider and stops before adding Anthropic SDK calls to non-Anthropic code. For Claude work, it detects the project language, then directs the agent to the matching reference files for Python, TypeScript, Java, Go, Ruby, C#, PHP or raw HTTP. It tells the agent to use the official Anthropic SDK when one is supported and to verify SDK details against the bundled references or current official sources rather than relying on remembered API shapes.

The folder also contains task-specific instructions for model and SDK migration, prompt audits, cost analysis, eval construction, preserved-thinking migration and related Claude API work. Those flows can inspect project files, propose or apply code changes when the request permits them, run local commands or scripts, and contact Anthropic services when the selected workflow calls for live API or documentation access. The resulting work depends on the requested task, such as Claude API code, a migration, an audit report, an eval setup or measured migration results.

Install

/plugin marketplace add anthropics/skills

Read the security check below before you run it. The repository's own instructions win if they differ.

Security check

What the skill's files do, from reading them. A flag that nobody has checked yet says "Not checked", never "No".

Runs scripts
Yes
Uses the network
Yes
Installs hooks
No
Asks for secrets
Yes

The folder includes four script files, and its workflows can run scripts and local commands. It also contains instructions that make network requests. Some workflows require Anthropic credentials; the preserved-thinking probe reads an API key or auth token from environment variables and sends requests to the Claude API. The scan found no installed hooks. Before installing, read the scripts and the workflow you plan to use, check which files and commands it may touch, and use a test repository when evaluating changes.

Scan details: Ships 4 script files: shared/evals/report/build-report-lite.mjs, shared/evals/report/runner-scaffold.mjs, shared/preserved-thinking-migration/drop_block_probe.py, shared/preserved-thinking-migration/prefix_diff.py. SKILL.md mentions an API key, token or secret variable. SKILL.md instructions include web requests or URLs to fetch. skills/claude-api/shared/preserved-thinking-migration/drop_block_probe.py makes network requests. skills/claude-api/shared/preserved-thinking-migration/drop_block_probe.py reads a key, token or secret from the environment. Checked at commit 8a1541c4a3ff.

Use cases

  • Build or modify code that calls the Claude API.
  • Check current Anthropic SDK patterns before writing integration code.
  • Migrate Claude API code, prompts or SDK usage.
  • Work with streaming, tools, caching, MCP or managed agents.

Limitations

  • It is for Claude and Anthropic SDK work, not other named model providers.
  • It does not generate Claude Agent SDK code.
  • Some workflows need network access and Anthropic credentials.
  • Language-specific API details must come from the bundled references or official sources.

Sources