Summary

Flyto-core provides a debuggable execution engine for AI agents and automation workflows, addressing the challenges of tracing, debugging, and replaying complex tasks. It offers a unified platform for diverse operations, from browser automation and API interactions to data processing and DevOps, enabling faster development cycles and enhanced reliability for engineers and researchers.

How It Works

The engine operates on a modular, MCP-native architecture, allowing AI agents to leverage its extensive library of 412 modules as tools. Workflows are defined in YAML or Python, with a core emphasis on debuggability. Key features include detailed execution tracing, step-level replayability, breakpoints, and data lineage, which significantly simplify debugging compared to traditional shell scripts or monolithic Python applications. This approach allows for granular re-execution of failed steps while preserving context, accelerating iteration and troubleshooting.

Quick Start & Requirements

• Installation:
  • Core engine: pip install flyto-core
  • Browser automation: pip install flyto-core[browser]
  • Browser setup: playwright install chromium (one-time)
• Prerequisites: Playwright (for browser modules), Chromium.
• Links:
  • Recipes documentation: docs/RECIPES.md
  • Module Specification: (Link not directly provided, but implied by "Module Authors" section)
  • Desktop GUI: flyto2.com

Highlighted Details

• Extensive Module Catalog: Features 412 modules across 78 categories, including browser automation (browser.*), flow control (flow.*), data manipulation (array.*, string.*, object.*), API integrations (api.*), image processing (image.*), and DevOps tools (docker.*, k8s.*).
• Debuggability & Replay: Offers full execution traces, per-step timing, input/output inspection, and the ability to replay workflows from any specific step.
• Unified Workflow Engine: Integrates browser automation, API calls, file I/O, data parsing, and more within a single, consistent framework.
• Multiple Interfaces: Supports CLI execution, an MCP server for AI agent integration, a RESTful HTTP API for remote execution, and a Python API for programmatic use.

Maintenance & Community

Contributions are welcomed, with guidelines provided in CONTRIBUTING.md. Security vulnerabilities should be reported to security@flyto.dev as per SECURITY.md. No specific community channels (e.g., Discord, Slack) are listed in the README.

Licensing & Compatibility

The project is licensed under the Apache License 2.0, making it free for both personal and commercial use without copyleft restrictions.

Limitations & Caveats

The README does not explicitly state limitations or known issues. The primary focus is on the core engine and CLI, with a separate URL provided for the desktop GUI.