Overview

xixo catalogs files and records that already live somewhere else. You attach the places they live, such as buckets, directories, mailboxes, calendars, feeds, wikis, and repositories. xixo reads each one and builds a single index over all of them. You can search that index, ask questions of it, and export from it. The bytes stay in their original places. xixo stores where each thing lives and what its analysis found.

Two interfaces reach the same catalog. The browser app uses GraphQL, and the MCP endpoint serves a model. Both accept the same access token, which masks issues. Each interface calls the domain layer directly, and neither is built on the other.

Key concepts

Feeds	the record for each file, note, address, tag, and content type, and the edges between them
References	where a feed’s bytes live, and how xixo notices a change to them
Adding things	uploads, notes, fetched pages, snapshots, and syncs, and how xixo chooses where a file is stored
Analysis	how xixo reads a feed: the steps, its text, what a model described, and the summary
Resources	the places xixo reads from, writes to, and sends requests to
Search	the index, and how lexical and semantic results are combined
Agents and MCP	the four tools, and the agent loop that calls them
Tenants	isolation between tenants, and signing in through masks
Security	each threat xixo defends against, and the defense

Start with the quickstart. The reference pages are generated from the code, or checked against it, so they match the running version.