AI Marketing Automation

AI-Marketing MCP Playbooks

Copy ready-to-run prompts for using Zeover from Claude, Cursor, Codex, or any MCP-aware AI agent. Start with setup, then audit, fix, generate, and operate your AI visibility loop.

44 MCP tools live SSE transport Z-API-Key header auth

Onboard with your AI

01

Ask your AI what Zeover does

Brand-new visitor. Wants to understand the product before paying for anything.

paste to Claude, OpenAI Codex, or any agent
$ What does Zeover do? Walk me through how it would help a B2B SaaS company that wants to show up when buyers ask ChatGPT for category recommendations.

Result: Plain-language overview, no MCP calls yet.

02

Get your account + API key

You're ready to install the MCP. You need to create an account and grab an API key.

paste to Claude, OpenAI Codex, or any agent
$ Open https://zeover.com/register, sign up free, then go to Settings -> API Keys and create one. Copy the key. Run the install command in scenario 3 with your key.

Result: Account + API key in hand.

03

Install the Zeover MCP

Wire Zeover into Claude Desktop, Claude Code CLI, Cursor, or any MCP-aware agent.

paste to Claude, OpenAI Codex, or any agent
$ claude mcp add --transport sse zeover https://api.zeover.com/api/v1/mcp/sse --header "Z-API-Key: YOUR_KEY"

Result: MCP server registered. Restart your AI client and you have 44 Zeover tools available.

Set up your brand

04

Create your brand

Spin up your first brand profile so Zeover knows what to optimize for.

paste to Claude, OpenAI Codex, or any agent
$ Using Zeover, create a new brand for me called "Example.com" at https://example.com. The brand sells AI-powered customer support to B2B SaaS companies. Set up the basics from the website.
# agent
Creating a new brand called 'Example.com'
Attaching Example.com as the brand's domain
Reading the auto-extracted onboarding profile

Result: Brand created, domain attached, onboarding auto-extracted from the site.

05

Review what Zeover learned about you

Auto-extraction may need editing. See what Zeover inferred from your site.

paste to Claude, OpenAI Codex, or any agent
$ Show me the onboarding profile Zeover generated for Example.com. List the target audience, selling points, and competitors it inferred. Highlight anything that looks wrong.
# agent
Loading the onboarding profile for Example.com

Result: Structured onboarding data. Edit anything that's off via the dashboard or by asking the agent to refine it.

06

Define your target queries

Choose the questions buyers ask AI engines that should put Example.com in the answer.

paste to Claude, OpenAI Codex, or any agent
$ For Example.com, add these target queries for benchmarking: "best AI customer support for SaaS", "top customer support AI tools 2026", "AI agents for SaaS helpdesk".
# agent
Adding 'best AI customer support for SaaS' as a tracked query
Adding 'top customer support AI tools 2026' as a tracked query
Adding 'AI agents for SaaS helpdesk' as a tracked query
Listing every tracked query to confirm

Result: Three queries staged. Each gets benchmarked across ChatGPT, Gemini, Grok, Claude, Perplexity.

Diagnose

07

Run a GEO audit

Get a baseline AI-readability score for your site.

paste to Claude, OpenAI Codex, or any agent
$ Run a Zeover audit on Example.com and summarize the failing checks by severity. Group them so I can see what to fix first.
# agent
Pulling the latest GEO audit for Example.com
Listing every crawled page for context

Result: Audit findings grouped Critical / Important / Low / Info, with remediation hints.

08

Drill into a specific page

One page is dragging the domain score. Find out why.

paste to Claude, OpenAI Codex, or any agent
$ Which Example.com page has the worst Zeover score and what specifically is failing on it? Show me the metric scope so I can see the offending content.
# agent
Listing all pages ranked by Zeover score
Loading the full analysis for the lowest-scoring page

Result: Per-metric failures with the exact text or markup that triggered each.

09

Where am I ranking in AI search?

See current position across every AI engine for every target query.

paste to Claude, OpenAI Codex, or any agent
$ For Example.com, show me my current rank position across every AI engine for every target query. Group by engine and flag any query where I'm not in the top 5.
# agent
Fetching the latest benchmark results across every AI engine

Result: Position matrix: rows are queries, columns are engines, cells are ranks 1-10 (or 'not ranked').

10

Who is AI recommending instead of me?

Find the brands AI cites when it should be citing you.

paste to Claude, OpenAI Codex, or any agent
$ For Example.com, list the top 5 competitor brands that AI engines cite ahead of me. Show which queries they win on. Tell me what content gives them an edge.
# agent
Identifying the top brands AI engines cite for your queries
Cross-referencing the benchmark rankings to see who wins where

Result: Competitor leaderboard + the queries each one wins.

Fix

11

Generate and publish llms.txt

Most sites are missing llms.txt entirely. This is the single highest-impact fix.

paste to Claude, OpenAI Codex, or any agent
$ Generate an llms.txt for Example.com based on the audit findings. Validate every referenced URL with an HTTP HEAD check. Show me the file and the validation report. Once it looks good, I'll publish it to my CDN.
# agent
Generating an llms.txt for Example.com and validating every referenced URL

Result: Validated llms.txt content. URLs verified live. Copy + paste to /llms.txt on your site.

12

Strip AI artifacts from content

Em-dashes, zero-width characters, mojibake. AI engines downweight these as machine-generated.

paste to Claude, OpenAI Codex, or any agent
$ Take the body content of Example.com/blog/why-ai-support and run it through Zeover's artifact remover. Return the cleaned version highlighted with diffs.
# agent
Scrubbing AI-tell artifacts (em-dashes, zero-width chars, mojibake) from the content

Result: Cleaned text + a diff of every removed artifact.

13

Reprocess your site after fixes

You shipped a wave of fixes. Confirm the score moved.

paste to Claude, OpenAI Codex, or any agent
$ I just published changes to Example.com. Trigger a fresh Zeover reprocess and tell me the new domain score and which checks now pass that didn't last week.
# agent
Triggering a fresh site re-crawl of Example.com
Pulling the new audit report once the crawl completes

Result: Domain rescanned, before/after score diff, list of newly-passing checks.

Generate

14

Write a citation-ready blog post

Long-form content built for AI engines to cite, in your brand voice, with social cross-posts.

paste to Claude, OpenAI Codex, or any agent
$ Generate a Zeover blog post for Example.com titled "How AI-powered customer support cuts ticket resolution time in half". Include three headline variants and a LinkedIn cross-post.
# agent
Drafting a long-form blog post in your brand voice
Watching generation status until the draft is ready

Result: Full blog draft (long-form, citation-structured) + 3 headlines + a LinkedIn post + an X thread starter.

15

Draft a press release

Funding round / feature launch / partnership announcement. Wire-ready format with cross-posts.

paste to Claude, OpenAI Codex, or any agent
$ Generate a press release announcing that Example.com just closed an $8M Series A led by Sequoia, with quotes from our CEO. Include LinkedIn and X variants.
# agent
Drafting a wire-ready press release with CEO quotes and cross-post variants

Result: Wire-format press release + supporting social posts.

16

Five social posts at once

Multi-channel campaign from one prompt. Brand-locked, channel-tuned.

paste to Claude, OpenAI Codex, or any agent
$ For Example.com, generate one LinkedIn post, one X post (under 280 chars), one Instagram caption, one TikTok caption, and one Facebook post about our new "Multi-language Support" feature. All in our brand voice.
# agent
Drafting a LinkedIn post tuned for executive feeds
Drafting an X post under 280 characters
Drafting Instagram and TikTok captions for the launch

Result: 5 channel-specific drafts. Each runs through brand-boilerplate enforcement + AI artifact removal.

17

App Store + Play Store descriptions

Need both iOS and Android store listings written for discovery + conversion.

paste to Claude, OpenAI Codex, or any agent
$ Example.com is launching iOS and Android apps. Generate the App Store description (under 3,999 chars) and the Play Store description (under 3,000 chars). Focus on AI customer support for B2B SaaS.
# agent
Drafting an App Store listing tuned to iOS discovery
Drafting a Play Store listing tuned to Android discovery

Result: Two store-ready descriptions tuned for each marketplace's search.

18

Optimize a LinkedIn profile

Founder/executive LinkedIn profile rewritten so AI engines cite them for category authority.

paste to Claude, OpenAI Codex, or any agent
$ Here's our CEO's current LinkedIn About section: [paste]. Rewrite it so AI engines will surface her as an authority on AI customer support. Keep it under 2,000 characters.
# agent
Rewriting the LinkedIn About section to position authority on the brand's category

Result: Rewritten About section, authority-positioned, brand-aligned.

19

Fill out a long form automatically

Trade show form, partner application, funding portal. Auto-fill with your brand profile.

paste to Claude, OpenAI Codex, or any agent
$ Here are the fields on this YC partnership application: [paste fields]. Use Example.com's brand profile to fill them all in.
# agent
Auto-filling every form field from Example.com's brand profile

Result: Filled answers, ready to paste into the form.

Operate

20

Tell ChatGPT/Gemini/Claude my content changed

You shipped new pages. Don't wait weeks for the AI training cycle to catch up.

paste to Claude, OpenAI Codex, or any agent
$ I just published 4 new blog posts to Example.com. Submit LLM update notifications for each so ChatGPT, Gemini, and Claude pick up the changes faster than the next training cycle.
# agent
Submitting LLM-update notifications to ChatGPT for each new post
Submitting LLM-update notifications to Gemini for each new post
Submitting LLM-update notifications to Claude for each new post

Result: Updates queued to each AI engine's indexing channel. Acknowledgement IDs returned.