# ClipCut: tickets from a Snoffle report (2026-10-31)

For a dev team. Each ticket says what to change, why, where, and how you know it is done. Do them in order; the first ones move the score most.

## Why this matters

Snoffle could not find your price on the page, so it cannot say what each sale is worth. Put the price on the page in plain text; buyers and their AI tools need it too. What is in the way: people asking Google or an AI assistant for this do not find you yet, so few get the chance to buy.

Your score stays at 40 or below while buyers do not find it and nobody mentions it yet, however good the page is. It lifts when buyers' searches find you and other sites start to mention you. The page fixes still count: they are what turns the first visitors into users.

Score today: 9/100 (AI can read it 25, found 0, mentioned 0).

## Tickets

### T1. Serve a real sitemap.xml
- Why: Snoffle saw /sitemap.xml not found (HTTP 404).
- Where: `public/sitemap.xml`
- Done when: curl -s https://clipcut.broke2builtai.com/sitemap.xml  → XML starting with <?xml, not HTML

<details><summary>The exact change</summary>

Goal: Give crawlers the list of public pages.
Where: public/sitemap.xml. Static files live in the folder your host serves as-is (public/ in most setups).
Depends on: nothing; it can be done on its own.
Risk: Low: adds or replaces one file. Rollback: delete the file or restore the old one.
Create `public/sitemap.xml` listing every public page (at least the home page):
```xml
<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url><loc>https://clipcut.broke2builtai.com/</loc></url>
</urlset>
```

</details>

### T2. Serve a real robots.txt that welcomes AI assistants
- Why: Snoffle saw robots.txt allows AI agents.
- Where: `public/robots.txt`
- Done when: curl -s https://clipcut.broke2builtai.com/robots.txt  → starts with "User-agent", not "<!doctype html>"

<details><summary>The exact change</summary>

Goal: Let search engines and AI crawlers read the site, and point them at the sitemap.
Where: public/robots.txt. Static files live in the folder your host serves as-is (public/ in most setups).
Depends on: step 1 (Serve a real sitemap.xml).
Risk: Low: adds or replaces one file. Rollback: delete the file or restore the old one.
Create `public/robots.txt` with exactly:
```
User-agent: *
Allow: /

User-agent: GPTBot
Allow: /
User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: Perplexity-User
Allow: /
User-agent: Google-Extended
Allow: /
User-agent: Applebot-Extended
Allow: /

Sitemap: https://clipcut.broke2builtai.com/sitemap.xml
```

</details>

### T3. Say the same price everywhere
- Why: Snoffle saw No price or "free" on the page.
- Where: the page or file named in the change
- Done when: Search the page source and each file for "$": every price matches.

<details><summary>The exact change</summary>

Goal: Make every place that states a price agree, so no assistant quotes the wrong one.
Where: The home page template or component: find it by searching the repo for the exact text Snoffle quoted above.
Depends on: nothing; it can be done on its own.
Risk: Medium: changes what visitors see. Rollback: revert the commit.
Pick one price and use it, word for word, on the page, in llms.txt, in the JSON-LD offers block and in any agent file (agent.json, openapi.json, SKILL.md). Remove old or test prices.

</details>

### T4. Add /llms.txt, the page AI assistants read first
- Why: Snoffle saw /llms.txt not found (HTTP 404).
- Where: `public/llms.txt`
- Done when: curl -s https://clipcut.broke2builtai.com/llms.txt  → starts with "# ClipCut"

<details><summary>The exact change</summary>

Goal: Give AI assistants a short plain-text summary they read first.
Where: public/llms.txt. Static files live in the folder your host serves as-is (public/ in most setups).
Depends on: step 3 (Say the same price everywhere).
Risk: Low: adds or replaces one file. Rollback: delete the file or restore the old one.
Create `public/llms.txt`:
```
# ClipCut

> The clip tool made only for standup comedians: upload a set, get your best jokes back as captioned clips. Price: [fill in: the price you settle in step 3].

## Links
- [Home](https://clipcut.broke2builtai.com/): what it is and how to start
- [Pricing](https://clipcut.broke2builtai.com/#pricing): the exact price
```
Use the same price and wording as the page.

</details>

### T5. Tell AI assistants your price in the format they read (JSON-LD)
- Why: Snoffle saw No JSON-LD on the page.
- Where: the page or file named in the change
- Done when: curl -s https://clipcut.broke2builtai.com/ | grep -c 'application/ld+json'  → 1 or more

<details><summary>The exact change</summary>

Goal: State the name, description and price in the structured format AI tools parse.
Where: the template that renders the home page's <head> (search the repo for "<title" or "<head").
Depends on: step 3 (Say the same price everywhere).
Risk: Low: one tag in the <head>. Rollback: restore the old tag.
In the <head> of the home page HTML (not injected by JavaScript), add:
```html
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "SoftwareApplication",
  "name": "ClipCut",
  "url": "https://clipcut.broke2builtai.com/",
  "description": "The clip tool made only for standup comedians: upload a set, get your best jokes back as captioned clips.",
  "applicationCategory": "WebApplication",
  "offers": { "@type": "Offer", "price": "[fill in: the price you settle in step 3]", "priceCurrency": "USD" }
}
</script>
```
Use "Product" instead of "SoftwareApplication" for a physical product. The price must match the one shown on the page.

</details>

### T6. Give agents a machine-readable way in
- Why: Snoffle saw No agent card, OpenAPI file or live API link.
- Where: `public/.well-known/agent.json`
- Done when: curl -s https://clipcut.broke2builtai.com/.well-known/agent.json  → JSON starting with {

<details><summary>The exact change</summary>

Goal: Give agents a machine-readable entry point.
Where: public/.well-known/agent.json. Static files live in the folder your host serves as-is (public/ in most setups).
Depends on: step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Low: adds or replaces one file. Rollback: delete the file or restore the old one.
Serve `public/.well-known/agent.json` as JSON (not your homepage):
```json
{ "name": "ClipCut", "description": "The clip tool made only for standup comedians: upload a set, get your best jokes back as captioned clips.", "url": "https://clipcut.broke2builtai.com/", "documentation": "https://clipcut.broke2builtai.com/llms.txt" }
```

</details>

### T7. Give AI agents one working link to follow
- Why: Snoffle saw No agent links to follow.
- Where: the page or file named in the change
- Done when: curl -s -o /dev/null -w "%{http_code}" <that link>  → 200

<details><summary>The exact change</summary>

Goal: Make every link an agent might follow load.
Where: The home page template or component: find it by searching the repo for the exact text Snoffle quoted above.
Depends on: step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Medium: changes what visitors see. Rollback: revert the commit.
Snoffle found no link for agents to follow. In llms.txt and on the page, link one path that works today: your API, your docs, or https://clipcut.broke2builtai.com/.well-known/agent.json. If you have none of those, link the page that says how to buy.

</details>

### T8. Point agents at your docs, API or MCP
- Why: Snoffle saw No API, OpenAPI, MCP, SKILL.md or agent card is named.
- Where: the page or file named in the change
- Done when: The links load and are in the raw HTML.

<details><summary>The exact change</summary>

Goal: Point agents at the docs, API or MCP server.
Where: The home page template or component: find it by searching the repo for the exact text Snoffle quoted above.
Depends on: step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Medium: changes what visitors see. Rollback: revert the commit.
Add a short "For AI agents" line on the page and in llms.txt linking your docs, API or MCP server (whichever exist).

</details>

### T9. Let an agent pay, or say plainly how to buy
- Why: Snoffle saw No payment method named.
- Where: the page or file named in the change
- Done when: llms.txt says how to pay in one line.

<details><summary>The exact change</summary>

Goal: Tell an agent exactly how to pay.
Where: Static files live in the folder your host serves as-is (public/ in most setups).
Depends on: step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Low: adds or replaces one file. Rollback: delete the file or restore the old one.
If an agent should be able to buy on its own, offer a machine payment path (an x402 endpoint, or a listing on a marketplace agents can pay on). If not, state the exact human checkout steps in llms.txt.

</details>

### T10. Name the product in the page title
- Why: Snoffle saw Title: "ClipCut" (7 chars).
- Where: the page or file named in the change
- Done when: curl -s https://clipcut.broke2builtai.com/ | grep -o '<title>[^<]*'

<details><summary>The exact change</summary>

Goal: Name the product and what it does in the page title.
Where: the template that renders the home page's <head> (search the repo for "<title" or "<head").
Depends on: nothing; it can be done on its own.
Risk: Low: one tag in the <head>. Rollback: restore the old tag.
Set <title> to "ClipCut: what it does in five words" (under 60 characters).

</details>

### T11. One clear action, as a plain link or form
- Why: Snoffle saw No button or link with an action verb.
- Where: the page or file named in the change
- Done when: The action works with JavaScript disabled.

<details><summary>The exact change</summary>

Goal: Make the main action work as a plain link or form.
Snoffle saw: The main action only runs in JavaScript.
Where: The home page template or component: find it by searching the repo for the exact text Snoffle quoted above.
Depends on: nothing; it can be done on its own.
Risk: Medium: changes what visitors see. Rollback: revert the commit.
Make the main action ("Start", "Buy", "Try") a real <a href> or <form>, not a JavaScript-only button.

</details>

### T12. Other sites mention you
- Why: Snoffle saw Google finds no page on another site that mentions clipcut.broke2builtai.com. To buyers and AI tools, nobody has used it yet.
- Where: the page or file named in the change
- Done when: Re-check on Snoffle.

<details><summary>The exact change</summary>

Goal: Fix what Snoffle found.
Where: The home page template or component: find it by searching the repo for the exact text Snoffle quoted above.
Depends on: nothing; it can be done on its own.
Risk: Medium: changes what visitors see. Rollback: revert the commit.
Get mentioned where your buyers already look: a launch post in the communities you belong to, a listing in the directories for your kind of product, and your first users saying it worked, in public

</details>

### T13. Make the page readable without JavaScript
- Why: Snoffle saw 0 characters of text before any script runs.
- Where: the page or file named in the change
- Done when: curl -s https://clipcut.broke2builtai.com/ | grep -i '<h1'  → your headline

<details><summary>The exact change</summary>

Goal: Put the headline, description, price and action in the HTML the server sends.
Where: the build config (render mode, prerendering), and the template that renders the home page's <head> (search the repo for "<title" or "<head") for preloads.
Depends on: nothing; it can be done on its own.
Risk: Medium: render or loading changes can break pages. Build and open the site locally before deploying. Rollback: revert the commit.
AI assistants do not run your JavaScript. Pre-render or server-render the home page so its headline, description, price and call to action are in the raw HTML (Lovable/Vite: add prerendering or a static fallback).

</details>

### T14. Expose one read-only endpoint that answers JSON
- Why: Snoffle saw No read-only endpoint was advertised to probe.
- Where: the page or file named in the change
- Done when: curl -s <endpoint>  → JSON

<details><summary>The exact change</summary>

Goal: Let an agent try the product through one read-only link.
Where: Your server routes (search for an existing GET handler and copy its shape).
Depends on: nothing; it can be done on its own.
Risk: Medium: a new public route. Keep it read-only, with no private data. Rollback: remove the route.
If your product has an API, link one safe GET endpoint (e.g. a status or search route) from llms.txt and the page, returning JSON. If it has no API, skip this step.

</details>

### T15. Let someone start without an account
- Why: Snoffle saw Nothing says an agent can start without an account.
- Where: the page or file named in the change
- Done when: A logged-out visitor can see it working.

<details><summary>The exact change</summary>

Goal: Let a visitor see it working before signing up.
Where: A new route beside the home page, in the same pages or routes folder; link it from the header or footer component.
Depends on: nothing; it can be done on its own.
Risk: Low: a new page and links to it. Rollback: remove the page and the links.
Give one way to see the product working before sign-up: an example, a demo or a free first step.

</details>

### T16. Be the answer when someone asks "Find me something that can take the recording of my standup set and cut out the best jokes as short captioned clips for TikTok and Instagram."
- Why: Snoffle saw Not in 9 results. Top: https://www.heygen.com/tool/long-video-to-short-video, https://www.heygen.com/en-in/tool/long-video-to-short-video, https://exemplary.ai/ai-clips.
- Where: the page or file named in the change
- Done when: Search Google for "Find me something that can take the recording of my standup set and cut out the best jokes as short captioned clips for TikTok and Instagram." and ask an AI assistant the same; re-check on Snoffle in a few days (indexing takes time).

<details><summary>The exact change</summary>

Goal: Have a page that answers the question buyers ask, so search and assistants surface it.
Snoffle saw: Google: not in the top 10 (top: youtube.com, reddit.com, facebook.com).
Where: A new route beside the home page, in the same pages or routes folder; link it from the header or footer component.
Depends on: step 1 (Serve a real sitemap.xml); step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Low: a new page and links to it. Rollback: remove the page and the links.
Add a page (or a section on your home page) whose first two sentences directly answer "Find me something that can take the recording of my standup set and cut out the best jokes as short captioned clips for TikTok and Instagram." and name ClipCut, with the price and how to start. Keep it in the raw HTML, link it from llms.txt and sitemap.xml, and give it a title that uses those words.

</details>

### T17. Be the answer when someone asks "I have an hour of comedy footage. Is there a cheap tool that automatically picks the funniest moments and adds subtitles so I can post them?"
- Why: Snoffle saw Not in 10 results. Top: https://www.heygen.com/blog/best-ai-tools-repurpose-long-videos-shorts-reels, https://www.vozo.ai/blogs/ai-tools-for-youtube-shorts, https://www.ssemble.com/tools/ai-clip-maker.
- Where: the page or file named in the change
- Done when: Search Google for "I have an hour of comedy footage. Is there a cheap tool that automatically picks the funniest moments and adds subtitles so I can post them?" and ask an AI assistant the same; re-check on Snoffle in a few days (indexing takes time).

<details><summary>The exact change</summary>

Goal: Have a page that answers the question buyers ask, so search and assistants surface it.
Snoffle saw: Google: not in the top 10 (top: reddit.com, apps.apple.com, choppity.com).
Where: A new route beside the home page, in the same pages or routes folder; link it from the header or footer component.
Depends on: step 1 (Serve a real sitemap.xml); step 4 (Add /llms.txt, the page AI assistants read first).
Risk: Low: a new page and links to it. Rollback: remove the page and the links.
Add a page (or a section on your home page) whose first two sentences directly answer "I have an hour of comedy footage. Is there a cheap tool that automatically picks the funniest moments and adds subtitles so I can post them?" and name ClipCut, with the price and how to start. Keep it in the raw HTML, link it from llms.txt and sitemap.xml, and give it a title that uses those words.

</details>

### T18. Be found when someone searches the name "ClipCut"
- Why: Snoffle saw Not in 9 results. Top: https://www.trustpilot.com/review/www.clipcut.pro, https://play.google.com/store/apps/details?id=medo.clipcut, https://github.com/akjadhav/ClipCut.
- Where: the page or file named in the change
- Done when: Search Google for "ClipCut" in a private window; re-check on Snoffle in a few days (indexing takes time).

<details><summary>The exact change</summary>

Goal: Be found when someone searches the product's name.
Human only. Don't attempt this step: it needs accounts or places outside the code. List it for your human with what follows.
Put "ClipCut" in the page <title>, the first heading and the first sentence, spelled the same everywhere. List ClipCut under that exact name where your buyers look (Product Hunt, a GitHub repo or README, the directories for your kind of product) and link each one back to https://clipcut.broke2builtai.com/. If the name is shared with something bigger, add a short qualifier everywhere (for example "ClipCut for <what it does>").

</details>

## Not code: what lifts the rest of the score

- Get mentioned where your buyers already look: a launch post in the communitie…
- People asking for this do not find you
- Searching your name does not find you

## Rules while you work

- Make only the change each task describes; keep the look, copy and behaviour otherwise.
- Never put an API key, token or password in any file.
- Do not change your prices while applying these fixes; a price change is a business decision, not a fix.

When it is deployed, re-check the same URL with Snoffle and compare the numbers.
