Image: GitHub
Why it mattersAn agency can now run a full SEO audit inside Codex CLI. The audits cover indexing, schema, Google AI Overviews and ChatGPT visibility.
Running an SEO audit used to mean opening eight tools in eight tabs. A new project called Codex SEO puts the checks inside OpenAI's Codex CLI. The author is a developer who publishes on GitHub as zhtyyx. The repository was first opened on 27 October 2026. Three days later it has 106 stars, 25 forks and 418 commits. It is under an MIT licence.
The author says Codex SEO is a port of a project called Claude-SEO. Claude-SEO was written by Agrici Daniel for Anthropic's Claude Code. The README says the methodology and the audit scripts come from the parent project. Both projects stay open source.
What one Codex session can check
The README lists five families of checks. Technical SEO covers crawlability, indexing, sitemaps, redirects, page speed and Core Web Vitals. Content quality covers the E-E-A-T framework, thin pages and topic coverage. Structured data covers Schema.org validation and generation. AI search covers Google AI Overviews, ChatGPT and Perplexity compatibility. Local and ecommerce cover Google Business Profile, reviews and product pages.
The project comes with 26 skills and 19 specialist roles, by the author's count. A role is a persona the agent takes on for a run. A technical SEO specialist reads one slice of a site, a local SEO specialist reads another. Each run writes a report with a short action list.
How a reader would use it
The install is one line in a terminal. After that the user runs Codex and names the audit they want. The output is a Markdown report. For an agency, that is the format a client reads. For a solo site owner, it is the format they paste back into the model. Then they can ask for help with the fix.
The author lists Google Business Profile, reviews and product-page coverage as separate audits. That is unusual for an open-source SEO tool. Most focus on the technical layer and stop. The AI search audits are the newest part of the method. A traditional SEO audit from a year ago does not cover them.
What it costs the user
Everything runs inside the Codex session. The user pays for the Codex calls that read the site and write the report. There is no hosted service and no account. The repository has 25 forks and seven open pull requests after three days. The author takes that as a sign that other SEO practitioners are reading the code.
The project is new. The 106 stars are a total, not a weekly gain, because the repo is three days old. The parent Claude-SEO project is older and well known in the Claude Code skills community. For a reader who already uses Codex every day, the cost of trying this is one install command. The reader who still uses a dashboard tool keeps that too.
Source
zhtyyx/codex-seo, on GitHub.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.