tauri_skill

Overview

This skill packages practical guidance for building secure, high-performance Tauri desktop apps that combine a Rust backend with a web frontend. It focuses on minimizing attack surface through least-privilege capabilities, strict IPC validation, strong CSP, and secure updater patterns. Use it to design, test, and harden Tauri applications with production-ready defaults.

How this skill works

The skill inspects typical Tauri attack surfaces and provides concrete patterns: capability scoping, content security policies, validated IPC command implementations, origin checks, async performance patterns, and secure updater configuration. It includes example code for Rust commands, validation logic, capability JSON, CSP entries, and lifecycle management so you can apply and test secure defaults quickly.

When to use it

• Starting a new Tauri project and choosing secure defaults
• Hardening an existing Tauri app against IPC, filesystem, or updater risks
• Designing scoped capability/permission models for desktop features
• Implementing or reviewing secure auto-update and cryptographic signing
• Writing tests to validate path access, origin checks, and input validation

Best practices

• Grant minimal capabilities; scope filesystem and shell access to allowlists only
• Validate and sanitize all IPC inputs using typed validators and canonicalize filesystem paths
• Use a restrictive CSP and freeze prototypes to prevent prototype pollution and XSS escalation
• Run async commands for I/O-heavy work and avoid blocking the main thread
• Use origin verification for sensitive commands and check window URL origins at runtime
• Protect updater endpoints with signed releases and verify public keys before install

Example use cases

• Implement a read_file command that canonicalizes and enforces app-directory scope
• Create capability profiles (default vs admin) to separate user permissions
• Add CSP entries limiting connect-src to required APIs and blocking frames/objects
• Design tests that assert path traversal is rejected and async commands return expected results
• Configure the updater to check signed releases and perform silent downloads with explicit user consent