- Home
- Skills
- Mobile Development
- sage chat app
More screenshots
Works with the AI Agents you already use
sage chat app
Scaffold and tailor private, multi-platform AI chat apps for Android and macOS using free models and local tools.
$10
sage chat app
Example session with this skill installed
Build a chat app named 'Cipher' with package com.studio.cipher. Include the core and code features, but omit google and sync. Put it in ~/Projects/cipher-app.
- Read your context and instructions
- Compiled the sage chat app
Cipher is ready
Built: core + code at ~/Projects/cipher-app. ./gradlew :shared:jvmTest :androidApp:assembleDebug :desktopApp:compileKotlin: 188 tests pass.
To start chatting:
- Get a free key at openrouter.ai/keys (or aistudio.google.com/apikey for Gemini) and enter it in Settings → API keys, on each device.
- Ensure Python 3 is installed on your Mac to enable the
codefeature for data analysis and charts. - Nothing else to set up as
googleandsyncmodules were excluded.
Next steps:
- Define your custom system prompt in
domain/SystemPrompt.kt. - Replace the placeholder logo in
ui/components/AppLogo.ktand the Android launcher icons. - Add your first custom tool by following the patterns in
references/architecture.md.
Later sessions: the generated CLAUDE.md holds the architecture rules, so the next session continues the same way.
Project Structure
| Module | Responsibility |
|---|---|
:shared | Core logic, Room KMP database, agent orchestration, and Compose UI |
:androidApp | Android-specific entry point and launcher configuration |
:desktopApp | macOS-specific entry point and menu bar integration |
:scripts | Scaffolding and installation utilities |
Feature Configuration
// Generated feature flags in shared/build.gradle.kts
val featureCode = true
val featureGoogle = false
val featureSync = false
val featureDictation = true
Connects securely to your tools. The creator never sees your data.
About this skill
Development of cross-platform AI chat applications is often bogged down by the complexity of Kotlin Multiplatform (KMP) configuration and the high cost of model APIs. This skill provides a production-ready scaffold for building private, feature-rich chat apps for Android and macOS that run entirely on free tiers and local models.
What it does
- Multi-platform scaffolding creates a unified Kotlin codebase for Android and macOS with shared UI and logic.
- Cost-effective routing implements smart model selection using free tiers from OpenRouter, Google Gemini, and local Ollama instances.
- Advanced agentic workflows enables multi-step planning, automated review cycles, and memory across chat sessions.
- Tool integration provides built-in support for MCP servers, web search via Tavily, and Python code execution.
- Privacy-first architecture ensures all chat history remains in an on-device database with optional E2EE Wi-Fi sync between devices.
How it works
- Configuration defines the app name, package ID, and optional feature modules like Google connectors or local sync.
- Scaffolding runs a specialized Python script to generate the project structure and wire up the KMP dependencies.
- Verification executes a comprehensive suite of ~210 tests to ensure the environment and core logic are stable.
- Tailoring allows for customizing system prompts, UI themes, and adding domain-specific tools or model providers.
Frameworks & tools
Kotlin Multiplatform, Compose Multiplatform, Room KMP, Koin DI, OpenRouter, Google Gemini, Ollama, and Tavily.
Why this beats prompting it yourself
Setting up a stable KMP project with AGP 9, KSP, and Compose is notoriously difficult to get right by hand. This skill uses a battle-tested template that resolves complex version conflicts and provides pre-built handling for streaming, artifacts, and local tool execution that simple prompts cannot reliably architect.
Use cases
- Build a private, local-first chat client for interacting with sensitive data via Ollama.
- Create a customized internal AI assistant for a team that integrates with Gmail and Drive.
- Launch a free-to-operate mobile chat app using OpenRouter's free-tier model routing.
- Develop a research tool that generates cited reports using web search and multi-step planning.
Known limitations
Requires a development environment with Python 3, JDK 21, and Android SDK. macOS features like MCP stdio and shell commands require a Mac. Sync requires both devices on the same local network.
Requires JDK 21, Android SDK, and Python 3. macOS features require a Mac. Features like Google connectors require manual OAuth client setup. Sync requires a shared local Wi-Fi network.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 16 days ago
Requires
Permissions
Allowed Hosts