Skip to main content
Reliant is designed with privacy as a core principle. Your code and conversations stay on your machine—we don’t have servers receiving your data.

​
Local-First Architecture

All data stays on your machine:
Data TypeStorage Location
Config~/.reliant/config.yaml (all platforms)
Conversations~/Library/Application Support/reliant (macOS), ~/.local/share/reliant (Linux)
API keys~/Library/Application Support/reliant (macOS), ~/.local/share/reliant (Linux)
Workflows.reliant/workflows/ in your project
Presets.reliant/presets/ in your project
Memories~/.reliant/reliant.md (global), reliant.md (project)
Worktrees~/.reliant/worktrees/ by default
Your codeOn your machine

​
What Goes to AI Providers

When you chat with Reliant, messages are sent directly to your chosen AI provider (Anthropic, OpenAI, etc.). This includes:
  • Your messages and prompts
  • Relevant code context (files you reference or that tools read)
  • Tool call requests and results
  • System prompts from workflows/presets
You control which provider receives your data by choosing your API provider in Settings → AI.

​
Provider Privacy Policies

Each provider has different data handling policies:
ProviderData RetentionTraining Usage
Anthropic30 days (API)Not used for training
OpenAI30 days (API)Not used for training (API)
GoogleVariesCheck their policy
Tip: Review your provider’s privacy policy. API usage typically has better privacy guarantees than consumer products.

​
Project Isolation

Each project has isolated data:
  • Separate chat history
  • Separate workflow configurations
  • Separate memory files (reliant.md)
Worktrees share the parent repository’s .reliant/ directory but have isolated conversations.

​
Offline Mode

Reliant works offline for:
  • Viewing past conversations
  • Editing workflows
  • File operations
  • Local LLMs

​
Telemetry

By default, Reliant does collect basic telemetry to understand what features you are using. This is not attached to any user information, and is only used for product feature decisions. If you opt-in to crash reporting, only:
  • Crash stack traces (no code content)
  • Basic system info (OS version, Reliant version)
are sent to help us fix bugs. You can opt out of both under Settings > Privacy.

​
Public Docs Site Analytics

The public docs site (docs.reliantlabs.io) is separate from the desktop app. The Mintlify docs experience includes browser-side analytics integrations for documentation usage. Apollo remains enabled on the public docs site. Statsig is also supported for anonymous docs analytics when it is explicitly enabled in the runtime configuration for a deployment. When Statsig is enabled, it collects anonymous documentation usage events such as:
  • Page views
  • Outbound link clicks
  • Docs search opens and submissions
The Statsig docs integration stores:
  • An anonymous visitor ID in localStorage
  • A session ID in sessionStorage
These docs site analytics do not include your local code, project files, or desktop chat history. Statsig stays disabled by default for local development and preview environments unless it is explicitly turned on for testing.

​
Enterprise Considerations

For organizations with strict security requirements:
  • Self-hosted providers: Use Azure OpenAI, AWS Bedrock, or local models
  • Air-gapped environments: Deploy with local LLM servers
  • Audit logging: Conversations are stored locally and can be reviewed
  • SSO/SAML: Coming soon

​
Deleting Your Data

​
Clear Conversations

In the app: Settings → Data → Clear Chat History

​
Complete Removal

# Remove user configuration (all platforms)
rm -rf ~/.reliant/

# macOS - Remove internal data and logs
rm -rf ~/Library/Application\ Support/reliant
rm -rf ~/Library/Logs/Reliant

# Linux - Remove internal data and logs
rm -rf ~/.local/share/reliant
rm -rf ~/.local/state/reliant

# Windows - Remove internal data
del /s /q %APPDATA%\reliant
Your code and project files are never modified by this process.

​
Security Practices

​
Code Modifications

Reliant shows you exactly what changes will be made before applying them. In manual approval mode, every edit requires your confirmation.

​
Tool Execution

Shell commands and file operations are:
  • Logged in the conversation
  • Visible in the UI before execution
  • Controllable via approval modes
Questions? Email security@reliantlabs.io for security-related inquiries.