Summary

Syll is a self-hosted AI companion runtime offering a web UI, chat channels, proactive rituals, editable markdown skills, recorded workflows, and a desktop ghost. It addresses the need for a proactive, context-aware AI assistant that integrates deeply with a user's digital environment, managing unfinished tasks and personal memories.

How It Works

Syll utilizes an event-driven architecture with an in-process MessageBus. Channels (web, CLI, Telegram, etc.) ingest input and publish to the bus. The AgentLoop orchestrates LLM calls, context assembly from workspace files and skills, tool dispatch (including GUI automation), and response generation. Proactive rituals and scheduled jobs are handled by a CronService, enabling autonomous actions.

Quick Start & Requirements

• Python: 3.11+ required.
• Installation: Setup environment (conda or venv), then pip install syll, syll onboard, syll wake.
• Model Config: LLM credentials (OpenRouter, OpenAI, etc.) in ~/.syll/config.json or web UI; local models supported.
• Dependencies: PyQt6/PyQt6-WebEngine for desktop ghost. macOS requires Accessibility/Screen Recording permissions. Docker support available.
• Links: Project page, Demos.

Highlighted Details

• MCP Tooling: Integrates external processes (stdio, SSE, HTTP) as tools with enhanced security and templates for Playwright, Blender, Godot.
• GUI Automation: Records desktop sessions into editable video/step traces for reusable workflows.
• Layered Memory: Notebook-like organization of long-term notes, daily fragments, and activity heatmaps.
• Proactive Rituals: Scheduled tasks and timely prompts with configurable silence.
• Desktop Ghost: Optional, always-on-top visual mascot mirroring Syll's state.

Maintenance & Community

Described as an "early public alpha" with ongoing development and a roadmap for features like multi-modal input. Contributions via PRs are welcomed. No specific community channels or direct contributor links are provided.

Licensing & Compatibility

Licensed under MIT, designed for readability and ease of use, generally permitting commercial use and integration into closed-source projects.

Limitations & Caveats

As an early alpha, Syll is under active development with pending features. The desktop ghost requires specific macOS permissions. Configuration involves managing LLM API keys and setting up communication channels.