Canonry

Canonry is an open-source, self-hosted platform from Canonry (formerly AI NYC) for tracking how AI answer engines such as ChatGPT, Claude, Gemini and Perplexity mention and cite a business. It combines that evidence with search, analytics and server traffic data, and exposes it through a web UI, CLI, REST API, MCP tools and webhooks. Project data is stored in SQLite and users bring their own provider API keys.

What It Is

Canonry is an agent-first answer engine optimization (AEO) operating platform. You run it on your own computer or server, add the queries that matter to a project, and schedule checks across AI providers. It records brand mentions and citations over time, shows which answers cite competitors but not your site, and lets you open the exact answers and URLs behind each result. It also runs bounded site health scans that produce a Page Health baseline.

How the Evidence and Actions Fit Together

The dashboard and a connected agent read the same project API. Evidence sources include Google Search Console, Bing Webmaster Tools, Google Analytics 4, Google Business Profile, server traffic from Cloudflare, Vercel, Cloud Run and WordPress, Common Crawl backlinks queried locally with DuckDB, and DataForSEO Labs keyword and ranking data. On the action side, Canonry supports WordPress publishing, JSON-LD structured data, and sitemap and URL submissions to Google, Bing and IndexNow. Scheduled data syncs and checks let you compare results after changes, and webhook alerts can go to Discord, Slack or other systems. The site notes that timeline comparisons show patterns and do not prove cause and effect.

Connecting Your Agent

Canonry does not replace an agent; it supplies evidence, tools and approval checks. Agents can use CLI commands with JSON output, a versioned REST API with an OpenAPI document, an MCP adapter, signed webhooks, or native Codex and Claude Code plugins. The site advises choosing either the native plugin or standalone MCP, not both. A bundled agent called Aero is also included.

Setup Path

Install with Homebrew, npm or Docker (Node 22+), then run bootstrap or init and start the server, and open the local setup page to scan a site. Provider keys for Gemini, OpenAI, Claude, Perplexity, Meta Muse, OpenRouter, Exa, Parallel or any OpenAI-compatible local model are configured separately. Canonry is single-tenant, so unrelated teams should use separate instances. Canonry 5.0 and later are MIT licensed.