MARTECHSIGNAL

Growth Lab Review

Open-source skills that run SEO and Xiaohongshu growth loops in Claude Code and Codex

Agent Skills · Open Source · OPEN SOURCE

Overview

Growth Lab is an agent skill pack that turns Claude Code or OpenAI Codex into a growth operator. You clone the repo, open the folder in your agent, and describe the growth outcome you want. The agent reads your product (a codebase, a prototype, or just a URL), researches what people actually search for, creates SEO pages aimed at those queries, publishes them, pings IndexNow, then reads the performance data and decides what to do next. The pitch is that context stops leaking between five disconnected tools. One workspace holds the product, the research, the output, and the results. Skills carry the method, and files carry the memory. Each capability is an observe-act-review loop with its own persistent memory, and the next run reads the previous results before it starts. The SEO loop maps user scenarios to live SERP research and generates pages that answer real queries. In the team's own test run, new pages got indexed within 1-2 days and page impressions and clicks both rose about 10x on a 7-day average. Average CTR dropped 50%, which they treated as input for the next iteration rather than a failure. The second loop targets Xiaohongshu: it collects high-performing posts in your niche, replicates the winning structure, writes the copy, generates images, runs a compliance check, and reviews the results. Their best test post drew 4,000+ likes and saves. Actual publishing on Xiaohongshu stays manual by design. Setup is git clone plus a conversation. An onboarding skill audits what's missing: API keys, browser sessions, third-party clients. You fix gaps in plain language instead of editing a config file. Secrets and cookies stay outside the workspace and never enter memory. The repo is Apache 2.0 licensed, has about 308 stars, and was pushed within days of this writing. The honest caveats: the project is roughly a week old. Only two loops actually work, and all the performance numbers come from a single run on the authors' own site. You need paid access to Claude Code or Codex, plus comfort working in a terminal with an agent. If your channels are X, LinkedIn, or TikTok, there's no loop for those yet. Who it's for: technical founders and small teams already living in a coding agent who want the growth work done rather than another dashboard. Compared with Surfer or Semrush, it writes and publishes the page instead of handing you recommendations. Compared with doing it manually, the research-to-publish cycle becomes one conversation. Compared with most agent skill packs, which teach the agent marketing knowledge, this one runs the loop end to end, memory included.

Screenshot of the Growth Lab homepage
Growth Lab homepage, captured September 2026. Vendor page shown as a dated reference capture; all site content belongs to its owner.

AI Capabilities

  • SEO page growth loop: scenario research, SERP analysis, page creation, IndexNow publishing
  • Xiaohongshu loop: viral post collection, structure replication, copy, images, compliance check
  • Persistent per-loop memory that feeds real results into the next run
  • Product understanding from a repo, prototype, URL, or notes
  • Natural-language configuration audit via the onboarding skill

Key Integrations

Claude CodeOpenAI CodexIndexNowBing Webmaster Tools

Pricing

Growth Lab is free to self-host under the Apache-2.0 licence.

Free, Apache 2.0. Self-hosted workspace; runs inside Claude Code or OpenAI Codex (both paid).

Review notes

Researched from public documentation, the source repository, and vendor materials. Not a hands-on test.

Growth Lab turns Claude Code or Codex into a growth operator: describe the outcome, and the agent reads your product from a codebase, prototype, or URL, researches real search queries, creates SEO pages around them, publishes, pings IndexNow, then reads the performance data and decides what to do next. It is a loop rather than one-shot generation, and the research and results share one workspace, so context stops leaking between tools.

Publishing from an agent means you own the pipeline: hosting, deploys, and content review on your side, which is a real commitment. The project is young at 308 stars, and the results ceiling is your model's judgment, so weak pages can ship at loop speed if you skip review. The Xiaohongshu growth loops are a rare China-market angle in Western tooling.

Verdict

Promising for teams ready to run self-hosted SEO loops with agent review. Everyone else should wait for maturity.

Pros and cons

ProsCons
Apache-2.0 licence with free self-hosting No hands-on test - this assessment is based on vendor documentation and the public repository
AI capabilities: SEO page growth loop: scenario research, SERP analysis, page creation, IndexNow publishing
Established community (1,999 GitHub stars)

Building your martech shortlist?

The weekly newsletter: one tool teardown, one workflow, no fluff. Free.

Subscribe

Frequently asked questions

What is Growth Lab?

Open-source skills that run SEO and Xiaohongshu growth loops in Claude Code and Codex. It ships with SEO page growth loop: scenario research, SERP analysis, page creation, IndexNow publishing, 1,999 GitHub stars. MartechSignal's review covers features, pricing, and how it compares to alternatives.

How much does Growth Lab cost?

Growth Lab is open source - Apache-2.0 licensed and free to self-host; the public repository carries 1,999 stars; native integrations cover Claude Code, OpenAI Codex, IndexNow. You pay in server time and maintenance, not licences.

Is Growth Lab a good Agent Skills tool in 2026?

Promising for teams ready to run self-hosted SEO loops with agent review. Everyone else should wait for maturity.

Similar Tools