AT PulseRead the daily briefRead
Oct 11 edition/Reporting & analysis
AgentsCodingBusiness

AgentsAutonomy & tool use

Julian Goldie video walks through SEO Machine, Craig Hewitt's open-source Claude Code workspace for SEO research and writing

SEO Machine, an MIT-licensed Claude Code workspace from Craig Hewitt, chains slash commands, specialist agents and brand context files to produce SEO research briefs and long-form drafts. A promotional walkthrough shows the workflow but offers no measured ranking or time-saving results.

30/100RisingHow it’s measured

YouTube video · 1.9K views · 30 likes

Reach blends public engagement (views, likes, comments, shares, stars) from the platforms where the story or its original source lives: YouTube, Hacker News, GitHub, Substack and similar. A video and the article written from its transcript share one score, taken from the video. Platforms with no data are left out rather than counted as zero, and the age of a story never lowers its score.

AT Pulse research deskPublished · 4 min read · AI-assisted analysis
THE CORE IDEAS4 TAKEAWAYS
01

SEO Machine started as an internal content pipeline Hewitt built for the podcast host Castos. It was released on GitHub in October 2025. It offers commands for research, writing, rewriting, optimization, auditing and performance review, along with about ten agents and 26 marketing skills. It connects to GA4, Search Console, DataForSEO and WordPress. [4] [5]

02

The workflow centres on research briefs. A brief covers primary and secondary keywords, search intent, target length, competitor coverage and content gaps. Context files for brand voice, style, SEO rules and internal links shape the output. Without a DataForSEO key, a brief contains only structure and strategy, with no live search data. [3] [4]

03

The video opens with a site going from zero to 1,300 daily clicks. That site is unnamed and was built with similar Claude-based systems, not with SEO Machine itself. The video credits its rankings mainly to branded search and backlinks. [2] [3]

04

Practical risks remain. One tracker records no updates since April 2026. Setup requires an Anthropic API account and Python dependencies, and DataForSEO costs extra. Under Google's spam policies, publishing AI-generated pages at scale without adding value for users counts as scaled content abuse. [4] [5] [8]

WHY IT MATTERS

the repository documents commands, analytics and publishing integrations, and scoring modules.

Read the full assessment

Implication: teams may speed brief and draft production, but rankings still hinge on brand demand, backlinks and editing, and mass publishing risks Google's scaled-content-abuse policy.

Executive brief

The video's headline is "zero to 1,300 clicks a day," but the same video says that unnamed site ranked well mainly because of branded search and backlinks (video, 06:53). It also says the site was built with "Claude-based SEO systems just like this one," not with SEO Machine itself. The tool is real: SEO Machine is an MIT-licensed Claude Code workspace by Craig Hewitt, first built for Castos (GitHub). It is widely starred, but one tracker reports no commits since April 2026 (Ry Walker). The Reddit post and the video come from Julian Goldie's channel and promote his course, consulting sessions and "Agent OS." No independent study of the tool's ranking results turned up.

What changed and event timeline

  1. Repository created

    SEO Machine went up on GitHub as an open-source version of an internal content pipeline Hewitt built for the podcast host Castos (;).

    rywalker.comgithub.com

  2. 7,121 stars, 968 forks

    Stars more than doubled in about three months, according to the same tracker ().

    rywalker.com

  3. GitHub Trending coverage

    An aggregator wrote it up as a trending Claude Code workspace for long-form SEO content (). Ry Walker logged 2,816 stars that month ().

    aitoolly.comrywalker.com

  4. Last recorded update

    Ry Walker flags the long gap since this update as a reason to keep watching the project ().

    rywalker.com

  5. Promotional video and Reddit post

    Goldie's channel posted a walkthrough presented by his "digital avatar" (), and the same pitch went to.

    youtube.comreddit.com

Capabilities and access

  • Commands: /research (plus SERP, gaps, trending, performance and topics variants), /write, /rewrite, /optimize, /analyze-existing, /publish-draft, landing-page commands, and /performance-review (GitHub).
  • Agents and skills: about ten agents (SEO Optimizer, Internal Linker, Keyword Mapper, Editor and others) and 26 marketing skills (same source).
  • Integrations: GA4, Search Console, DataForSEO, and WordPress with Yoast metadata (same source).
Read the full section
  • Commands: /research (plus SERP, gaps, trending, performance and topics variants), /write, /rewrite, /optimize, /analyze-existing, /publish-draft, landing-page commands, and /performance-review (GitHub).
  • Agents and skills: about ten agents (SEO Optimizer, Internal Linker, Keyword Mapper, Editor and others) and 26 marketing skills (same source).
  • Integrations: GA4, Search Console, DataForSEO, and WordPress with Yoast metadata (same source).
  • Requirements: Claude Code, an Anthropic API account and Python dependencies (GitHub). DataForSEO costs extra (Ry Walker).
  • Model version: neither source names a specific Claude model.

Technical analysis for researchers and developers

  • Architecture: slash commands run specialist sub-agents and five Python modules that score search intent, keywords, length relative to competitors, readability and overall quality (GitHub).
  • Context files: these files steer the output. The README treats getting them right as critical to quality.
  • Live data: without an API key, the brief only gives structure and strategy.
Read the full section
  • Architecture: slash commands run specialist sub-agents and five Python modules that score search intent, keywords, length relative to competitors, readability and overall quality (GitHub).
  • Context files: these files steer the output. They cover brand voice, style guide, SEO rules, target keywords, an internal-links map and competitor notes. The README treats getting them right as critical to quality. It also sets a 2,000-word minimum for articles.
  • Live data: without an API key, the brief only gives structure and strategy. Live search data needs DataForSEO (video, 03:15).
  • Evaluation: the reviewed sources contain no benchmark, ablation or ranking experiment.

Claims and evidence

  • This is a single demo with no timing data.
  • The unprompted research brief. In the video, Claude ran a full brief for "Claude SEO agent" before being asked for research (video, 01:57).
  • 1,300 visits a day. The site is unnamed, and the video credits brand search and links for the result (video, 06:35).
Read the full section
  • "Hours of research in minutes" (video, 00:12). This is a single demo with no timing data.
  • The unprompted research brief. In the video, Claude ran a full brief for "Claude SEO agent" before being asked for research (video, 01:57). This is anecdotal.
  • 1,300 visits a day. The site is unnamed, and the video credits brand search and links for the result (video, 06:35). It is uncorroborated and not attributable to SEO Machine.
  • Adoption numbers. Star and fork counts are confirmed on GitHub (7.5k stars, 985 forks at last check). Stars measure interest, not results.

Context and prior work

  • SEO Machine is one of several Claude Code SEO toolkits. Others include the claude-seo skills, which grade E-E-A-T signals and how likely a page is to be cited by AI search engines (claudeskills.info).
  • Ry Walker places SEO Machine on the content-production side. He says it lacks technical audits and AI-search optimization, and recommends pairing it with an audit tool (Ry Walker).
Read the full section
  • SEO Machine is one of several Claude Code SEO toolkits. Others include the claude-seo skills, which grade E-E-A-T signals and how likely a page is to be cited by AI search engines (claudeskills.info).
  • Ry Walker places SEO Machine on the content-production side. He says it lacks technical audits and AI-search optimization, and recommends pairing it with an audit tool (Ry Walker).

Limitations, safety and contested findings

  • Google policy: using generative AI to produce many pages "without adding value for users" counts as scaled content abuse under Google's spam policies.
  • Setup effort: the video calls setup "a couple of commands" (01:24), but the README lists API accounts and Python dependencies.
  • Third-party code: the video tells viewers to read any skill before running it (01:47).
Read the full section
  • Google policy: using generative AI to produce many pages "without adding value for users" counts as scaled content abuse under Google's spam policies. A tool built for volume publishing carries that risk.
  • Setup effort: the video calls setup "a couple of commands" (01:24), but the README lists API accounts and Python dependencies.
  • Third-party code: the video tells viewers to read any skill before running it (01:47).
  • Maintenance: development appears paused (Ry Walker).
  • Caption quality: the video's automatic captions may contain transcription errors.

Business and practitioner implications

  • Where it helps: teams whose bottleneck is turning out research briefs and drafts. The context files make brand voice consistent and easy to check.
  • Costs: budget for Anthropic API usage, DataForSEO, and human editing.
  • Measuring value: run your own A/B tests against hand-written briefs using Search Console data. The vendor and influencer traffic stories don't establish results.
Read the full section
  • Where it helps: teams whose bottleneck is turning out research briefs and drafts. The context files make brand voice consistent and easy to check.
  • Costs: budget for Anthropic API usage, DataForSEO, and human editing.
  • Measuring value: run your own A/B tests against hand-written briefs using Search Console data. The vendor and influencer traffic stories don't establish results.
  • Pick-up risk: pin a version or fork the repo, given the commit gap.
  • Ranking signals: the video itself puts the source of ranking gains in brand demand and backlinks, not AI-generated text (07:03).
FOLLOW THE EVIDENCE

The source trail.

Sources (8)
01

SEO Machine Claude Code Does Hours of SEO Research in Minutes

Feed excerpt only. A transcript of the video linked from this page is supplied (linked video auto captions; https://www.youtube.com/watch?v=MlMWXzb39ZQ); automatic text may contain errors.

reddit.com
02

Claude SEO Took This Site From 0 to 1,300 Clicks a Day

Related coverage; assess separately

reddit.com
03

Linked video: SEO Machine Claude Code Does Hours of SEO Research in Minutes

linked video auto captions

www.youtube.com
04

SEO Machine — GitHub repository

github.com
05

Ry Walker — SEO Machine research note

rywalker.com
06

AIToolly — SEO Machine (2026-03-08)

aitoolly.com
07

claude-seo seo-content skill

claudeskills.info
08

Google Search Central — Spam policies

developers.google.com
A LITTLE LESS NOISE. A LOT MORE CONTEXT.

Stay curious.
Follow the evidence.

Independent perspectives, the original sources, and room for the questions that don't have easy answers.

How we build the brief
Connect with us

Find us where you already read.