The problem
How it works
In a 29-minute walkthrough, Cody Schneider calls this marketing engineering: a coding agent doing the hands-on work between a marketing plan and its results. He argues the agent should read from a warehouse fed by a data pipeline instead of calling the Meta API directly, and lists the pieces: a server, storage, Postgres, scheduled jobs, auth and git. He installs the Graphed CLI to set these up from Claude Code, makes a square static ad with an image model, creates a Meta system-user token, uploads the ad and later switches ads off from the terminal. He then asks one question across Facebook, GA4 and Search Console that lands in a dashboard, turns people who engaged with a LinkedIn post into verified emails, finds keywords ranking on page two, writes and publishes a blog post with an SEO plugin, and drafts UGC ad scripts from complaints on Reddit.
Results and limitations
Marked reported: the video’s chapters show each step being done on camera, but it gives no outcomes such as ad performance, traffic or reply rates. Graphed is the presenter’s own company, so treat it as a vendor demo.
Before you try it
A Meta system-user token that never expires can spend money on your ad account, so keep it in a secret store, give it the fewest permissions and set spending limits in Ads Manager. Collecting LinkedIn engagers and enriching their emails raises platform-terms and privacy questions, so check the rules where you sell before you send anything. Read ads and posts before they go live. Results depend on your setup and were not independently reproduced by Top 200 Repos.
Tools and repositories
- Tools
- Claude Code · Graphed CLI · Meta Ads, GA4 and Search Console
- Prerequisites
- A Graphed account, a Claude Code plan and admin access to the ad and analytics accounts you connect.
Try it yourself: Ask one read-only question across your marketing data
Fill the fields to adapt the prompt. Empty fields stay as visible [placeholders].
Using the data in [Where the data lives], compare [Metric] for [Period] across [Sources]. Show a table by [Breakdown], flag anything that changed by more than [Change threshold], and suggest one change to test next week. Read only: do not change any campaign, budget or ad.Check two or three numbers against the ad platform’s own dashboard before acting on the table.
Related videos
About anthropics/claude-code
You Can Turn Obsidian Into JARVIS (Claude Code) Anyone Can Do This (opens in a new tab)
Walkthrough of claude-code by Maitis: the title names the project; the description links the project website.
Watch on YouTube (opens in a new tab)About anthropics/claude-code
The SECRET Claude Code Upgrade That Has Everyone Going Nuts (Full Guide) (opens in a new tab)
The title names the project, the description links the project website.
Watch on YouTube (opens in a new tab)Independent creators. Listing a video implies no endorsement, sponsorship or partnership.
Sources
- Marketing Engineering Crash Course with Claude Code (opens in a new tab)
First-person account · youtube.com · Published 14 Oct 2026 · Source checked 13 Nov 2026, 19:26 UTC
What this source supports
- Claude Code is set up with the Graphed CLI to run a marketing stack from chat
- Recommends a data pipeline and warehouse instead of the agent calling the Meta API directly
- Uploads a static ad to Facebook Ads and turns ads off from Claude Code using a system-user token
- One prompt across Meta, GA4 and Search Console into a dashboard; LinkedIn engagers to verified emails
- Keyword research, page-two keywords, a blog post published with an SEO plugin, and UGC scripts from Reddit pain points