Let your coding agent integrate Indx

Most search products give developers documentation to read. Indx gives your coding agent everything it needs to do the integration itself, and to verify its own work against a running search engine. You describe what you want searchable. The agent does the rest.

Don't read our docs. Give them to your agent.

A prompt like this is enough:

Add Indx search to this project. Products are in products.json;
name and brand searchable, category and price filterable.

Three pieces make that work.

An agent skill

The Indx skill is packaged, versioned knowledge of how to build with Indx: the C# and HTTP APIs, field configuration, filters, coverage tuning, and the version pitfalls an agent would otherwise fall into. One install line for Claude Code. It is open source.

Docs your agent can read

The entire documentation site ships as a single file at docs.indx.co/llms-full.txt. Using Cursor, Copilot, or an agent you built yourself? Point it there and it has the full picture.

A built-in MCP server

Every IndxCloudApi exposes read-only search tools over the Model Context Protocol, self-hosted servers included. Four tools: list datasets, describe a dataset with value hints so the agent filters on real values instead of guessing, search, and fetch a document. The server uses the same API keys as the REST API, and admins can turn it off.

This is the part that sets the workflow apart. The agent does not just write integration code that looks right. It searches the live dataset, sees actual results, and corrects itself until the integration is verified against a running engine.

The loop

The skill teaches. The agent builds. The MCP server verifies. You describe what should be searchable, and your agent takes it from prompt to working, tested search.

And if you build agents rather than with them, Indx suits the runtime too. Embedded via NuGet it runs in-process with sub-millisecond queries. No service, no cold start.