Open source · Self-hosted

The operating system for AEO

Canonry watches what ChatGPT, Claude, Gemini, and Perplexity say about you, joins it to your search and server evidence, and hands your agent the tools to fix what it finds.

View on GitHubView npm package

Latest release v4.192.0. Local SQLite. Your provider keys. Your infrastructure.

Canonry dashboard showing answer-engine mention trends across Claude, Gemini, OpenAI, and Perplexity
Install CanonryNode 22+
brew tap canonry/canonrybrew install canonry

Then run cnry init and cnry serve.

Any agent · Local workflow

Runs inside the agent you already use.

CLI, REST, MCP, webhooks, or a native Codex or Claude Code plugin. AUQ, a SaaS SEO and AEO agency, reviewed that surface from the outside.

Use the native plugin or standalone MCP, not both. Plugin guide

Connect your agentNative plugin
codex plugin marketplace add Canonry/canonrycodex plugin add canonry@canonry

First run cnry init --skip-skills --skip-mcp, then cnry start.

One typed system

One record. Four ways in.

Shared project recordCanonry
30+

CLI commands

cnry runs --tail --format json
118

REST endpoints

GET /api/v1/citations
48

MCP tools

tools/competitor_gap
6

Webhook events

citation.gained

Measurement context

One timeline. Three sources.

Align by date and URL. Patterns, not attribution. Server events catch the crawler and referral activity a browser analytics tag never records, and the full walkthrough covers the loop from first sweep to client report.

AI answersMentions · Citations
Google SearchImpressions · Clicks · CTR
Server eventsCrawlers · Fetches · Referrals
Compare byDate + URL
  1. 01

    Search demand

    Impressions · Clicks · CTR

    Google Search Console performance showing clicks, impressions, and click-through rate over 30 days
  2. 02

    Server activity

    Crawlers · Fetches · Referrals

    Server event report showing crawler activity, AI user fetches, AI referral sessions, and request evidence

The operating loop

Evidence in. Approval out.

  1. 01

    Connect evidence

  2. 02

    Run baseline

  3. 03

    Agent prepares

  4. 04

    You approve

Full operating surface

Observe. Act. Operate. Connect.

01

Observe

  • AI citationsGemini · ChatGPT · Claude · Perplexity · local LLMs
  • Server trafficCloudflare · Cloud Run · Vercel · WordPress
  • Search performanceGSCGA4Bing
  • Local AEOQueries · performance · lodging + booking gaps
  • Common Crawl backlinksMonthly graph · scheduled sync · DuckDB
02

Act

  • WordPress executionContent fixes
  • JSON-LD schemaStructured data
  • Indexing submissionsGoogle · Bing · IndexNow
03

Operate

  • Many clientsYAML · cnry apply
  • Scheduled syncsVisibility · traffic · Business Profile
  • Regression alertsWebhooks
  • Client reportscnry report <project>
04

Connect

  • Your own agentCLI · REST · webhooks
  • MCP adapterAny MCP-capable agent
  • Native pluginsCodex · Claude Code
  • AeroBuilt in · wakes after every run

Two-minute walkthrough

See the local workflow.

Canonry walkthrough video thumbnail reading CLI to AI answers, showing the CLI and local dashboard side by sideWatch on YouTube

Direct answers

Common questions about Canonry.

What is Canonry?

Canonry is the operating system for AEO: an agent first AEO + web analytics monitoring and execution platform that is open source and self-hosted. It tracks how AI answer engines mention and cite a site, joins that evidence with search, analytics, technical, and traffic data, and exposes the same system through a dashboard, CLI, REST API, MCP tools, and webhooks.

Does Canonry replace my agent?

No. Canonry is the layer underneath it. It supplies evidence, typed tools, budgets, and verification to the agent and local workflow you already use, through CLI, REST, MCP, webhooks, or a native Codex or Claude Code plugin. Aero ships as the built-in agent.

Where does Canonry store my data?

Canonry is self-hosted and single-tenant. By default, its project data lives in one SQLite file on your machine or server. You control the provider keys, backups, and deployment.

Which AI engines can Canonry monitor?

Canonry supports Gemini, OpenAI, Anthropic Claude, Perplexity, and OpenAI-compatible local models. A browser provider can also capture citations from supported consumer interfaces when the web experience differs from an API response.

How does my agent connect?

Codex and Claude Code can install the native Canonry plugin from the Canonry marketplace. The CLI remains the simplest universal path because every command can return JSON. Canonry also exposes a versioned REST API, an MCP adapter with toolkits that load on demand, and signed webhooks for event-driven workflows.

Is the source public?

Yes. The source, issues, architecture, and documentation are public in the Canonry GitHub repository. The npm package is published as @canonry/canonry under the Functional Source License, which converts to Apache 2.0.

Open source · Local-first

Stop guessing who cites you.

Install Canonry and run the whole loop yourself.

Get CanonryExplore open-source tooling
The Open Source Operating System for AEO | Canonry