# Get Maguyva wired into your agent

Canonical source: https://maguyva.ai/docs

License: text under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/); code samples under [MIT-0](https://opensource.org/license/mit-0); the Maguyva name, logos, brand assets and other code are not covered. Details: [maguyva.ai/license](https://maguyva.ai/license).

Your agent starts every session blind. Maguyva fixes that with codebase search, symbol lookup, and dependency context over MCP.

The shortest path is simple: start free, index a small representative repo set, connect your client, then ask one real question you already know how to verify.

## Four moves from blind agent to grounded answers.

The shortest path is still the best one: connect a small repo set, wire your client, then ask a question you already know how to verify.

- **Start free**: No credit card required. Use a small representative repo set instead of trying to index your whole company on day one.
- **Index a repo you already know**: Pick code you can sanity-check fast. That makes it obvious whether Maguyva is grounding the model better than a generic search pass.
- **Connect your MCP client**
- **Ask a real repo question**: Jump straight to symbols, callers, definitions, and dependencies — every result cited by file path, with line numbers on search hits — instead of hoping the model guessed right.

## Documentation Overview

- [Quickstart](https://maguyva.ai/docs/quickstart): From zero to a first verified answer in a few minutes
- [Installation](https://maguyva.ai/docs/installation): Wire Maguyva into your MCP client in a few minutes
- [How It Works](https://maguyva.ai/docs/how-it-works): See how indexing, search, symbols, and dependency graphs fit together
- [MCP API Reference](https://maguyva.ai/docs/mcp-api): Tool-by-tool reference for the Maguyva MCP surface
- [Compatibility](https://maguyva.ai/docs/compatibility): Supported languages, repository limits, and file coverage
- [Cookbook](https://maguyva.ai/docs/cookbook): Short recipes for install, search-before-edit, empty results, and multi-repo
- [Troubleshooting](https://maguyva.ai/docs/troubleshooting): Fix API key, bridge, and repository connection issues

## More resources

- [Language Guides](https://maguyva.ai/languages): See how Maguyva handles your language stack
- [Blog](https://maguyva.ai/blog): Architecture decisions and technical deep-dives
- [Pricing](https://maguyva.ai/pricing): Per-workspace pricing, starting free

## Validate on one question you can already grade.

Maguyva looks best when you use it against code you already understand. Start with one repo, ask one real question, inspect the file paths, and make the tool earn trust before it becomes part of your default loop.

- Question: What code handles login and session expiry?
- Check: Open the cited files, follow the symbols, and verify that the answer survives contact with the actual repo.
- Signal: Smaller context, fewer blind spots, better odds that the model edits the right file the first time.

### What you get back

- Grounded answers with file paths and line numbers.
- Symbol lookups, callers, definitions, and dependency paths in one surface.
- One remote MCP server that works across multiple clients instead of a separate workflow for every editor.