Skip to main content

Framework integration guides

Use these guides when wiring IterableData into AI frameworks or provider SDKs. Longer provider notes stay in the repository so llms.txt and GitHub remain the single source of truth; this page summarizes what each guide covers and links to the in-docs agent APIs.

In-docs guides (start here)​

GuideWhen to use
Building agentsDesigning an agent that detects formats, samples rows, and converts data
MCP serverExposing IterableData tools over Model Context Protocol (iterabledata[mcp])
Agent tools APICalling detect_format, read_sample, plan_conversion, and related tools from Python
Catalog APIMachine-readable format/capability metadata for planners
AI APINative providers (openai, anthropic, gemini, azure) and doc.generate

Provider and framework notes (GitHub)​

GuideSummary
AI frameworksLangChain, CrewAI, and AutoGen patterns for tools and memory
OpenAIOpenAI SDK usage with iterable.ai and redaction
Anthropic ClaudeClaude / Messages API wiring
Google GeminiGemini provider setup

Install AI extras with pip install 'iterabledata[ai]' (add [mcp] for the MCP server). Never commit API keys; use redact_for_llm() before cloud calls.

Extending formats outside core​

For a niche format that should not land in the main package, ship a small plugin package with the iterabledata.formats entry point. See Plugin system and the reference plugin walkthrough for a complete layout.