GoManzanas · Chrome Extension

Turn browser steps into polished guides — instantly.

GuideBot captures your clicks and screenshots, lets you annotate them, then uses AI to generate a clear, step-by-step guide you can share with anyone.

Publisher GoManzanas
Version 1.1
Contact extensions@gomanzanas.com
01

What GuideBot Does

GuideBot is a Chrome extension that records your browser actions and turns them into professional how-to guides — without leaving your browser.

📸

Capture Steps

Click the capture button as you work. GuideBot screenshots each step automatically and records the context of every action.

✏️

Annotate Screenshots

Draw circles, arrows, and numbered markers directly on screenshots to highlight exactly what the reader should focus on.

🤖

AI-Generated Guides

Send your captured steps to Claude, Gemini, or OpenRouter. The AI writes clear, numbered instructions from your actions.

🔒

Your Keys, Your Data

API keys are stored only on your device. Screenshots and steps never leave your browser unless you explicitly send them to an AI provider.

💬

Clarifying Questions

Toggle accuracy mode to let the AI flag ambiguous steps before writing — asking about hidden field values, unclear selections, or cut-off text — so the guide is based on facts, not guesses.

✍️

Edit & Refine

After generation, switch to Edit mode to tweak the markdown directly in the preview panel. Changes re-render instantly and are saved back to your session.


02

How It Works

From capturing your first click to a shareable guide in under a minute.

1
Open GuideBot from the toolbar

Click the GuideBot icon in your Chrome toolbar to open the panel. Add your AI provider and API key once in Settings — it's saved locally for future sessions.

2
Capture steps as you work

Click Capture Step at each point in your workflow. GuideBot takes a screenshot of the current tab and records the action. Repeat for every step you want to document.

3
Annotate your screenshots

After each capture, an editor opens so you can draw circles, drop numbered markers, or add arrows to call out the key elements. Annotations are baked into the screenshot.

4
Generate the guide

Name your guide, then click Generate Guide. GuideBot sends your annotated screenshots and step context to the AI provider you've selected.

5
Answer clarifying questions (optional)

If Ask clarifying questions is enabled in Settings, GuideBot first asks the AI to review your screenshots for anything ambiguous — a field with hidden text, an unclear menu selection, or cut-off content. Answer as many or as few as you like, then proceed with full context.

6
Preview, edit, and export

Review the finished guide in the preview tab. Click Edit to modify the markdown directly — changes re-render live and are saved to your session. When satisfied, copy the text or export to HTML, Markdown, Word, or PDF.

Tip
You can re-order or delete individual steps before generating. If a screenshot didn't capture the right moment, remove it and recapture — the session stays open until you clear it. After generation, use Edit in the preview to fine-tune any wording the AI got wrong.

03

Choose Your AI Provider

GuideBot works with the AI model you already use. Bring your own API key — GuideBot never proxies your requests through our servers.

Anthropic
Claude models
Google Gemini
Gemini models
OpenRouter
Any supported model

Setting up your API key

1

Open GuideBot and click the Settings button (gear icon).

2

Select your provider from the dropdown: Anthropic, Gemini, or OpenRouter.

3

Paste your API key and click Save. The key is stored in chrome.storage.local on your device only.

Key Security
Your API key is stored locally in Chrome's sandboxed storage and is only ever sent directly to the AI provider you've selected. GuideBot has no backend — API calls go from your browser straight to the provider's API endpoint.

Improving accuracy with clarifying questions

Enable the Ask clarifying questions toggle in Settings before clicking Generate. GuideBot will make a quick pre-pass to the AI — sending your screenshots and asking it to return a structured list of anything it can't determine from the image alone. Answer what you can, skip what you can't, and the final guide will incorporate your answers as ground truth.

When to use it
Turn on clarifying questions whenever steps involve text typed into fields that aren't visible in the screenshot, selections from long dropdowns, or multi-step actions on a single page. The pre-pass adds one extra API call but significantly reduces invented details in the output.
ProviderKey FormatWhere to Get One
Anthropic sk-ant-… console.anthropic.com
Google Gemini AIza… aistudio.google.com
OpenRouter sk-or-… openrouter.ai/keys

04

Privacy at a Glance

GuideBot is designed to keep your data in your hands. No GoManzanas servers. No telemetry. No data collection.

DataWhere It LivesWho Sees It
API keys chrome.storage.local on your device You only — sent directly to your chosen AI provider
Screenshots & step data In-memory during your session; chrome.storage.local between sessions You only — sent to AI provider only when you click Generate
Annotations Baked into the screenshot PNG on-device You only
Generated guide text & guide name chrome.storage.local between the generation and preview tab; updated if you edit in Edit mode You only — overwritten on next generation
Clarify mode preference chrome.storage.local on your device You only

For the full policy, see the GuideBot Privacy Policy.

Uninstalling
Removing GuideBot from Chrome permanently deletes all locally stored data — API keys, steps, and screenshots. There is nothing on any GoManzanas server to delete.

05

Chrome Permissions

GuideBot requests the minimum permissions needed to capture and annotate browser steps.

PermissionWhy GuideBot Needs It
activeTab Capture a screenshot of the tab you're currently viewing when you click Capture Step.
storage Persist your API keys and in-progress steps across browser sessions so you don't lose your work.
tabs Read the current tab's title and URL to include in the step context sent to the AI for accurate guide generation.
scripting Inject the capture listener into the active tab to detect click coordinates and page context for each recorded step.
No Host Permissions
GuideBot does not request broad host permissions like <all_urls>. The content script is injected only into the active tab at the moment you trigger a capture — not persistently into every page you visit.

06

Support & Contact

Questions, feedback, or bug reports — we're here to help.

ChannelDetails
Email extensions@gomanzanas.com
Publisher GoManzanas
Chrome Web Store Use the "Support" tab on the GuideBot store listing
API Key Issues
If guide generation fails, check that your API key is valid and has sufficient credits with your provider. GuideBot surfaces the raw error from the AI provider's API so you can diagnose the issue directly.