Sugar

Sugar is an autonomous agent that monitors a GitHub repository for labeled issues such as security alerts, Dependabot updates, or bug reports. When an issue appears, Sugar reads its description, uses an AI model to generate a code fix, runs the project's test suite and quality checks, and then opens a pull request that references the original issue for review.

The tool maintains both project‑specific and global memory stores, allowing it to recall decisions, preferences, error patterns, and coding guidelines across sessions. Memory can be accessed and updated by connected AI agents through the MCP interface, enabling contextual reasoning without manual copy‑pasting. A task queue drives the autonomous workflow, while semantic search retrieves relevant context by meaning rather than keyword matching.

Sugar is invoked via a command‑line interface; after installing with pipx, users initialize a project, configure which labels to watch, and run `sugar run`. The system supports global guidelines that apply to all projects and integrates with Claude Code and other MCP clients for seamless AI‑assisted development.