Getting started

Coding agent prompts

AI coding agents work best when they have accurate, current context about the APIs they’re integrating with. AssemblyAI’s API surface changes — model names, parameters, and SDK methods evolve — so agents that rely on training data alone will produce outdated code.

The methods below are ranked by how reliably they keep your agent’s output correct. Use all three for the best results, or start with project instructions if you only pick one.

Project instructions

Add this line to your project’s CLAUDE.md, .cursorrules, or equivalent agent instructions file:

Always fetch https://assemblyai.com/docs/llms.txt before writing AssemblyAI code.
The API has changed — do not rely on memorized parameter names.

This tells your agent to pull the latest documentation before generating code, which prevents hallucinated parameters and outdated model names.

Why this is the most effective method: Project instructions run on every prompt. The agent will check current docs before writing any AssemblyAI code, catching breaking changes automatically.

Additional context files

AssemblyAI publishes machine-readable documentation indexes:

  • llms.txt — Structured index of all documentation pages
  • llms-full.txt — Full concatenated documentation content

Filter by language to reduce token usage:

/docs/llms-full.txt?lang=python
/docs/llms-full.txt?lang=typescript

Exclude the API spec to focus on guides:

/docs/llms-full.txt?excludeSpec=true

MCP server

Connect the AssemblyAI docs MCP server to give your agent tool-based access to search and read documentation on demand.

Server URL:

https://mcp.assemblyai.com/docs

Connecting (any MCP client with Streamable HTTP transport)

Claude Code:

$claude mcp add assemblyai-docs --transport http https://mcp.assemblyai.com/docs

Claude Cowork:

Settings > Connectors > Scroll to the bottom > Click Connect [AssemblyAI Docs]

Cursor (.cursor/mcp.json):

1{
2  "mcpServers": {
3    "assemblyai-docs": {
4      "url": "https://mcp.assemblyai.com/docs"
5    }
6  }
7}

Any MCP client that supports Streamable HTTP transport can connect using the server URL above.

Available tools

The MCP server provides four tools:

  • search_docs — Search across all documentation pages
  • get_pages — Retrieve full content of specific pages
  • list_sections — Browse the documentation structure
  • get_api_reference — Get API endpoint details and schemas

Claude Code skill

The AssemblyAI skill gives Claude Code curated instructions and context for working with AssemblyAI. It covers the Python and JavaScript SDKs, streaming, voice agents, audio intelligence, and more.

Install the skill:

$claude install-skill https://github.com/AssemblyAI/assemblyai-skill

Works with 60+ AI coding agents via the universal skills format:

$npx skills add AssemblyAI/assemblyai-skill

Verify it’s installed:

$claude skill list

Tips for best results

  • Layer all three tools — Project instructions catch every prompt, the MCP server provides on-demand lookups, and the skill gives deep SDK context
  • Be specific about the SDK — Say “use the AssemblyAI Python SDK” or “use the JavaScript SDK” rather than “use AssemblyAI”
  • Reference the model — Specify universal-3-pro for pre-recorded or u3-rt-pro for streaming to avoid outdated model names
  • Set your API key as an environment variable — export ASSEMBLYAI_API_KEY=your_key so the agent can reference it naturally