Do I need Claude Code or Codex?

Yes for the work. The results app works in any browser.

Does it use my AI subscription?

Yes. Your agent does the writing and coding. Credits only pay for search data, AI visibility checks and verification.

Will it post on Reddit for me?

It finds the thread, drafts the reply and fills the box in your browser. You press post.

What if my Reddit account is new?

The audit tells you which subs you can post in yet, and gives you a plan to take part for real until you can.

Which site stacks work?

Next.js, Astro, Nuxt, SvelteKit, Hugo and static HTML at launch. Webflow, Framer and WordPress come next, through the browser.

Does it work for iOS apps?

Yes, on the App Store. Add your app's link: the listing audit and the review to-do list are free. Builder ships the fixes and gives your app its own site on shipfound.site; Growth adds screenshot A/B tests, custom product pages and the Apple Ads agent. You submit, publish and approve every change.

Changes are pull requests on a branch, and the agent never merges. Nothing leaves your machine except what you record with Shipfound.

Do I need to replace Google Analytics?

No. It runs alongside GA4 or PostHog and can import their history. It is cookieless by default; turn on Attribution mode for multi-day journeys and A/B tests, and it respects your consent banner.

I don't have much traffic. Can I still A/B test?

Sometimes. Before any test starts, Shipfound tells you how long it would take to get an answer at your traffic. If that is too long, it suggests a test that can finish, or ships the change and tracks it as a before-and-after.

Do you guarantee rankings?

No. We guarantee the work and the measurement, and the Growth plan stops charging if you're not named by day 90.

For founders who build in Claude Code and Codex

Your agent ships features. Now it ships customers.

Shipfound checks what you have access to, then fixes your site for Google and AI search, ships your blog and glossary as pull requests, gets you indexed and listed, and drafts the right replies. You approve. We verify every result.

Install in Claude Code Install in Codex

Free to install. The audit costs no credits.

free · about 3 min

$ claude plugin marketplace add shipfound/shipfound
$ claude plugin install shipfound@shipfound
# then, in a new Claude Code session in your repo:
> /shipfound:audit

Run the first two in a terminal. They work for Claude Code in the terminal, VS Code, Cursor and the desktop app. The first command gives you a link and a code: open it, check the code and choose Allow.

~/code/shipfound-site · claude

shipfound.coDAY 23 / 90

Named status

Named by Perplexity in 1 of 4 runs, day 23

ChatGPT0 of 4
Perplexity1 of 4
Google AI1 of 4
Claude0 of 4

Shipped this week

llms.txt and AI crawler rulesVerified
Listed on SaaSHubVerified
PR #12 Structured dataClaimed

Illustrative session and numbers. Our real ones go on the proof page from day 0.

02 · It starts with what you have

First, it checks your access.

Repo, Search Console, Bing, Gmail, Reddit, X, GitHub. Three minutes later you know which channels are open, which are blocked, and the one thing that unblocks each. Your plan only contains moves you can make.

/shipfound:audit

Access Card 5 open 1 caveat 2 blocked

RepoNext.js 15, content in content/, push accessOpen
Search ConsoleProperty verified, no sitemap submitted→ Submit /sitemap.xmlCaveat
BingSite not added→ Import the site from Search ConsoleBlocked
GmailConnected, sends from a company domainOpen
RedditHas an account to post from. Drafts come to you; you post themOpen
XLogged in, 380 followers, active this weekOpen
GitHubgh signed in, 2 awesome lists that fit and do not list you yetOpen
TrackingNo analytics or AI crawler beacon on the live site→ Your agent installs it by pull request: /shipfound:fixBlocked

Illustrative card. Yours lists all eleven areas the audit checks.

03 · What it ships

Work that lands in your repo and accounts, not in a dashboard.

Site fixes.

Metadata, structured data, llms.txt, AI crawler rules, internal links. One pull request per theme, with a preview.

/shipfound:fix

Content.

Glossary, answer pages, comparisons and alternatives, researched with live search data and checked before the PR.

/shipfound:write

Indexed.

Sitemaps on Google Search Console and Bing, IndexNow on every new page. Bing's index feeds ChatGPT search.

/shipfound:index

Listed.

The directories, review sites and marketplaces AI engines actually cite, filled in your browser.

/shipfound:list

Conversations.

Reddit, X, Quora and HN threads where buyers ask for what you sell, with replies drafted to each platform's rules.

/shipfound:reach

Inbox.

Personal notes to signups, your network, podcasts and listicle authors, saved as Gmail drafts.

/shipfound:reach

Measured.

One script and a crawler beacon. Every channel, conversion and dollar, with ChatGPT and Perplexity as channels of their own, traced to the shipped work that caused it.

/shipfound:analytics

Tested.

Your agent proposes the A/B test, writes the variant, runs it, calls it honestly and ships the winner.

/shipfound:test

3b · Every day

Day 1, the audit. Every day after, one command.

No list of commands to work through. Each day, one command in your repo reads what happened since your last session and what is still failing on your live site, then offers the one move worth making today, with what it gets you and its price.

Day 1/shipfound:auditChecks your access and starts the baseline.
Every day after/shipfound:growChecks what changed, then offers one move.

In Codex, say: "What should I do today?"

~/code/your-site · claudeExample

> /shipfound:grow

Since yesterday: 3 visits from ChatGPT, 212 AI crawler hits, 2 signups.

Due: PR #14 (llms.txt) merged 2 days ago. Verify it? 1 credit.

Test "pricing-headline": variant B wins. Close it and keep B?

Today: 2 site checks still failing on the live site: OG image on /pricing, FAQ structured data. One PR, 6 credits. Start?

Example brief. Yours is built from your own site, numbers and work in progress.

01
It looks before it offers.
What changed since your last session, what is due to verify, tests with a verdict, checks still failing on the live site, directories still to list on, the next page to write. Looking is free.
02
One move, never the same one twice running.
Only moves where it found concrete work, with names and numbers, what it gets you and why now. Decline one and it waits a week.
03
Nothing runs without your yes.
Anything over 5 credits states its price first. Every result says Claimed until our server verifies it.

04 · See your AI traffic

Know when ChatGPT sends you a customer.

Most analytics lump AI traffic into "referral" and never see AI crawlers at all. Shipfound shows GPTBot, PerplexityBot and ClaudeBot reading your new pages, the visits that follow from ChatGPT and Perplexity, and the signups they turn into, all traced back to the work that caused them. Free on every plan.

How do I know if AI crawlers are hitting my site?

1Day 0
PR merged
/glossary/llms-txt · GitHub
Verified
2Day 2
GPTBot crawl
Crawler beacon · 200 OK
3Day 9
Named in 2 of 4 runs
ChatGPT · visibility run
4Days 9 to 30
31
ChatGPT visits
tracking.js · AI Search
5Days 9 to 30
6
Signups
Goal: signup

Illustrative timeline for one shipped page.

4b · Experiments your agent runs

Your agent runs the A/B test too.

It reads your funnel, finds where visitors leave, and proposes three tests with the evidence behind each. Pick one and it writes the variant as a pull request. When the result is in, it tells you in plain words and opens the PR that ships the winner. If your traffic is too low to get an answer, it tells you before you start, not six weeks later.

exp-hero-headlineDay 21 of 21 · decision reached

Hypothesis

Leading with the outcome instead of the mechanism gets more visitors to start signing up. 62% of AI Search visitors leave the hero within 10 seconds.

A · control

B · variant

B: 94% chance to beat A, +18% signup starts

-3%0+18%+39%

95% credible interval for the uplift

Ship winner · PR #31Removes the flag and the losing code.

Illustrative experiment.

4c · For iOS apps

Building an app? It works the App Store too.

Add your App Store link and the same agent audits your listing, turns your reviews into a to-do list, and ships the fixes. It writes your app a site that Google and AI answers can find, sets up screenshot A/B tests and custom product pages for you to start, and runs Apple Ads with your approval on every change. You submit. We verify on the live App Store.

/shipfound:aso

Listing audit

Keyword field, subtitle, every language your storefront reads, and a description AI answers can quote. Checked against the ASO rules.

Free

Reviews as a to-do list

Your 1 and 2 star reviews grouped into themes, the most costly first, with what to fix and a reply for you to post.

Free

Fixes in your next version

New keywords, subtitles and languages, staged in the version you submit or opened as a PR. Verified once Apple releases it.

Builder

Your app's own site

Most apps have no website, so Google and ChatGPT find nothing. Your agent writes a page for each search on yourapp.shipfound.site; you publish it.

Builder

Screenshots and A/B tests

New captioned screenshots at App Store size, tested against your current ones in Apple's own A/B test. You press start.

Growth

Custom product pages

A product page for each audience, with its own screenshots and the search keywords it shows for, linked from your site.

Growth

Apple Ads agent

Reads your campaigns and proposes each change with its numbers: keywords that install, searches that only spend, bids, budget. Nothing is spent until you approve it.

Growth

Downloads by source

App Store search, browse, web and each site page's campaign, from App Store Connect, next to what your agent shipped.

Free

05 · Proof, not a score

One AI check lies. We ask four times.

Ask ChatGPT the same question twice and you get different brands. We ask every question 4 times on every engine, keep every answer, and label results as "named in 3 of 4 runs". Every action shows Verified only after our server checks it live.

What is the best way to track if ChatGPT mentions my product?

ChatGPT · "best tool to get my SaaS mentioned by AI search" named in 2 of 4 runs

Run 1Named

1. Northwind SEO
2. Lumen Rank
3. YourProduct

Run 2Not named

1. Lumen Rank
2. Acme Metrics
3. Northwind SEO

Run 3Named

1. YourProduct
2. Northwind SEO
3. Orbit AEO

Run 4Not named

1. Acme Metrics
2. Lumen Rank
3. Orbit AEO

Illustrative answers with made-up brand names. Real runs keep every answer, word for word.

06 · What it never does

Your name is on it. We act like it.

Never posts, publishes or sends without you.
Never merges code.
Never uses accounts you don't own.
No fake reviews.
No karma farming.
No bulk email.

07 · Pricing

Credits pay for data and checks. Your agent does the thinking.

Free

Access audit
One baseline visibility check: 10 questions, 4 engines, 4 runs each
20 credits
Analytics for 1 site, 10k events a month, 30-day history
App Store listing audit and review to-do list

Install free

Builder

$49/mo

300 credits a month
Weekly visibility re-check on 25 prompts, 3 competitors
The results app and a public proof page
Full analytics, 200k events a month, 13-month history
All attribution models, 3 concurrent experiments
App Store fixes, and your app's own site on shipfound.site

Start Builder

Growth

With the promise

$99/mo

700 credits a month
Everything in Builder
The 90-day named promise
1M events a month, 3 sites
Unlimited experiments, warehouse export
App Store A/B tests, custom product pages and the Apple Ads agent

Not named by day 90? You pay nothing more until you are. What "named" means

Start Growth

Credit packs: $15 per 100, never expire.Events over the plan: $10 per 500k. Tracking never stops.

Definition

"Named": Your product appears in at least 2 of 4 runs, on at least one engine, for at least one of your tracked questions.

Not named by day 90 on Growth, you pay nothing more until you are: up to 3 free months, as a credit, not a cash refund.

08 · FAQ

Questions founders ask first.

Install it. Run the audit. See what's open.

Install in Claude Code Install in Codex

Free to install. The audit costs no credits.

Your agent ships features. Now it ships customers.

First, it checks your access.

Work that lands in your repo and accounts, not in a dashboard.

Site fixes.

Content.

Indexed.

Listed.

Conversations.

Inbox.

Measured.

Tested.

Day 1, the audit. Every day after, one command.

It looks before it offers.

One move, never the same one twice running.

Nothing runs without your yes.

Know when ChatGPT sends you a customer.

Your agent runs the A/B test too.

Building an app? It works the App Store too.

Listing audit

Reviews as a to-do list

Fixes in your next version

Your app's own site

Screenshots and A/B tests

Custom product pages

Apple Ads agent

Downloads by source

One AI check lies. We ask four times.

Your name is on it. We act like it.

Credits pay for data and checks. Your agent does the thinking.

Free

Builder

Growth

Questions founders ask first.

01Do I need Claude Code or Codex?

02Does it use my AI subscription?

03Will it post on Reddit for me?

04What if my Reddit account is new?

05Which site stacks work?

06Does it work for iOS apps?

07Is my code safe?

08What does "named" mean?

09Do I need to replace Google Analytics?

10I don't have much traffic. Can I still A/B test?

11Do you guarantee rankings?

Install it. Run the audit. See what's open.