Summer access

Full access through August 31

Summer is hot, so Ghost Bro is open. Activate your trial by August 25 and your access runs until summer ends. Existing accounts are included; activating after August 25 uses the regular 7-day trial.

20% off until August 31 HOT20

Settings & Configuration

Learn how to customize GhostBro to fit your workflow, from selecting AI models to configuring privacy and security settings.

Access the Settings Panel by clicking the gear icon in the main control panel or using the Cmd + , (macOS) or Ctrl + , (Windows/Linux) shortcut. GhostBro's settings are organized into clear sections to help you manage your AI experience.

๐Ÿ”‘ AI & API Configuration
๐ŸŽ™๏ธ Transcription Settings
๐Ÿง  Smart Features
๐Ÿ›ก๏ธ Privacy & Security

AI & API Configuration

Flexible AI Providers

GhostBro is built to be provider-agnostic. Configure any service that is OpenAI SDK compatible:

  • Paid Services: Use your own keys for OpenAI, Anthropic, or Google Gemini.
  • Free & Open Source: Connect to providers like OpenRouter or Groq.
  • Local LLMs: Point GhostBro to a local server (e.g., Ollama, LM Studio) for 100% privacy and zero latency.

Note: All API keys are securely encrypted using Electron Safe Storage before being saved to your local machine.

Custom Models

Tailor your AI experience by adding specific models from any provider:

  • Model Identifiers: Define any model ID (e.g., gpt-5.4, claude-4-6-sonnet, or your fine-tuned local models).
  • Custom Endpoints: Configure custom base URLs for specialized or private API gateways.

Provider support matrix

GhostBro can use local tools or external providers depending on the workflow. Use this matrix to choose the route that fits your privacy, cost, and latency needs.

Provider route Used for External API key Best fit Privacy note
OpenAI-compatible chat Screen analysis, summaries, translations, prompt actions Usually yes Hosted chat models, custom gateways, or compatible local servers External requests follow the configured provider's policy unless the endpoint is local.
Anthropic Claude Chat, reasoning, summaries, writing help Yes Longer reasoning and careful text work Data sent to Anthropic is handled by Anthropic's policies.
Google Gemini Chat, screen-context analysis, summaries Yes Teams already using Google's AI stack Data sent to Gemini is handled by Google's provider terms.
OpenRouter or Groq OpenAI-compatible chat routes Yes Model choice, routing flexibility, or low-latency hosted models Requests go to the configured external provider.
Ollama, LM Studio, or local server OpenAI-compatible local model route No, unless your local gateway requires one Local/private use, testing, and predictable cost Model requests can stay on your machine or private network.
OpenAI or Deepgram transcription Speech-to-text transcription Yes Managed transcription speed, language support, and reliability Audio sent to the provider is subject to that provider's policies.
Local Whisper Speech-to-text transcription No Offline or privacy-sensitive transcription Audio can remain on the device when local transcription is selected.

Transcription Settings

๐ŸŽ™๏ธ

Universal Providers

Hybrid Approach: Use cloud providers (OpenAI/Deepgram) for speed or Local Whisper for 100% offline privacy.

๐ŸŒ

Context & Language

Select target languages and provide domain-specific prompts to improve keyword accuracy across any industry.

๐Ÿ”Š

Intelligent Routing

Capture from your Microphone or System Audio to bridge your digital and physical conversations.

Unlimited Skills & Custom Prompts

GhostBro isn't just for developersโ€”the sky is the limit. Transform your assistant into an expert in any domain by defining custom skills and prompts.

Domain Expertise

  • Beyond Coding: Create skills for Law, Medicine, Sales, Security, or Design.
  • Expert Personas: Instruct the AI to act as a "Senior Architect", "Creative Director", or "Legal Consultant".

Custom Prompts

  • Actionable Logic: Customize behavior for Solve, Analyze, or Brainstorm tasks tailored to your specific workflow.
  • Context Window Control: Fine-tune how much conversation history is analyzed for each prompt.

Markdown Notes

  • Speaker Notes: Upload .md documents to serve as formatted speaker notes during presentations.
  • Instant Toggle: Use Cmd + 7 to show or hide your notes in a discrete, semi-transparent overlay.

Privacy & Security

๐Ÿ‘ป Ghost Mode

Total invisibility. Enable Ghost Mode to hide the GhostBro window from screen shares, recordings, and screenshots.

๐Ÿ“‹ Clipboard Controls

Granular control over what gets copied. Choose to include AI insights, direct answers, or both.

Maintenance

Zero Friction Updates

Stay ahead with the latest features and model supports through manual or automatic checks.

Configuration Reset

Experimental settings gone wrong? Reset to factory defaults with a single click.

Onboarding

Need a refresher? Restart the guided tour to master all configuration possibilities.

Provider configuration FAQ

When should I use a local model?

Use a local model when privacy, offline work, or predictable provider costs matter more than the latest hosted model quality.

What happens to data sent to an external provider?

Data sent to OpenAI, Anthropic, Gemini, Deepgram, OpenRouter, Groq, or another configured provider is handled by that provider under its own policies and billing rules.

Where does Ghost Bro store provider keys?

The public documentation states that provider keys are encrypted on the user's machine with Electron Safe Storage.

Need Help?

Explore more documentation or reach out to our team if you're stuck.