Open-source agent tool Canonry released

Arber Xhindoli’s AI NYC account posted on X on May 19 that Canonry is available as an open-source tool for monitoring how AI answer engines cite websites and send traffic. Public package and repository pages show the software is positioned as an “agent-first” AEO, or answer engine optimization, platform built for both human analysts and automated workflows. The release materials published Tuesday point to a GitHub repository, an npm package and product documentation describing support for ChatGPT, Gemini, Claude and Perplexity. The project’s public docs say the software can track citations, run sweeps, store evidence and expose results through a web interface, CLI and API. ### What, exactly, did Canonry release on Tuesday? The GitHub repository for AINYC/canonry describes Canonry as “an agent first AEO monitoring and operating platform,” and the repo showed 233 commits, 30 stars and six forks when checked Tuesday. The latest visible commit was attributed to “arberx” about one hour earlier and referenced pinning OpenClaw and requiring Node 22.14. The npm package page showed `@ainyc/canonry` version 1.27.2 was published two hours before it was checked. (github.com) That page describes Canonry as software that tracks how AI answer engines “cite or omit your website” and says it is built so “AI agents and automation pipelines can operate it end-to-end without human intervention.” ### What does “agent-first” mean in Canonry’s own documentation? (github.com) The npm documentation says every Canonry capability is exposed through “a stable REST API and a machine-readable CLI.” The same page says an agent can install the package, configure providers, create projects, trigger visibility sweeps and act on the results from a terminal. Canonry’s public package docs say the web dashboard is optional and that “nothing requires it.” The package page also says the CLI and API cover “100% of functionality,” while setup can be done non-interactively through flags or environment variables for use in CI, containers or agent workflows. (npmjs.com) ### Which AI systems and workflows does it say it supports? AI NYC’s open-source tooling page says Canonry can “run agents, orchestrate workflows, monitor citations, and automate everything” across ChatGPT, Claude, Gemini and other systems from a web UI, CLI and API. (npmjs.com) The npm package page separately lists ChatGPT, Gemini, Claude, Perplexity “and others” as answer engines it can monitor. The package documentation gives a sample workflow in which a coding agent installs Canonry, runs `canonry bootstrap`, starts a background daemon, applies a YAML project spec, triggers a run and then inspects status, evidence and history in JSON. (npmjs.com) The docs say every command supports `--format json`, which allows agents to parse output directly. ### What else is included besides ranking or citation checks? (ainyc.ai) The repository history and related public pages show Canonry’s release materials include more than basic citation monitoring. A recent docs commit referenced “social media referral tracking to traffic section,” while the npmx package summary says the tool can watch AI engines crawl and refer traffic through server-log ingestion and includes built-in Google Search Console, GA4 and Bing Webmaster diagnostics. (npmjs.com) AI NYC’s tooling page places Canonry alongside other public components including `@ainyc/aeo-audit` and OpenClaw or Claude Code skills. That page says the broader toolkit is published with transparent scoring, CI-friendly output and documented workflows for auditing, fixing, validating and monitoring. ### Where can users verify or try it now? The public npm page says users can install the package with `npm install -g @ainyc/canonry`, then run `canonry init` and `canonry serve`, with the optional dashboard available on localhost port 4100. (github.com) The same page says headless setup is also available through `canonry bootstrap`. The next public updates are likely to appear in the AINYC/canonry GitHub repository, where commits were still landing Tuesday, and in the npm package feed, which showed a fresh publish for version 1.27.2. (ainyc.ai) (github.com) (npmjs.com)