agent-sh/agentsys
AI writes code. This automates everything else · 24 plugins · 49 agents · 44 skills · for Claude Code, OpenCode, Codex, Cursor, Kiro.
What it solves
AgentSys is a modular runtime and orchestration system designed to handle the complex operational overhead of AI-driven software development. While AI models can write code, AgentSys manages the surrounding lifecycle: task selection, branch management, code review, artifact cleanup, CI monitoring, and deployment, ensuring that AI agents follow structured pipelines rather than operating in isolation.
How it works
It functions as a marketplace and installer for a ecosystem of plugins, agents, and skills. The system orchestrates specialized agents—each with a single responsibility and specific model assignment—into phase-gated pipelines. This prevents agents from skipping critical steps (like testing or review) and maintains persistent state across sessions.
To optimize cost and quality, it uses a hybrid approach: deterministic tools (regex, AST analysis) handle detection, while LLMs are reserved for high-level judgment and synthesis. This structure allows lower-cost models (like Claude Sonnet) to achieve results comparable to high-tier models (like Claude Opus) by providing them with enriched context and structured prompts.
Who it’s for
It is built for developers using AI coding tools such as Claude Code, Codex CLI, OpenCode, Cursor, and Kiro who want to automate end-to-end software delivery pipelines.
Highlights
- End-to-End Workflows: Commands like
/next-taskautomate the entire flow from task discovery and planning to implementation, review, and PR creation. - Quality Gates: Integrated pre-ship checks including
/deslop(to remove AI artifacts) and/prepare-delivery(to validate test coverage and build status). - Durable Memory: The
/banthiscommand allows users to create persistent negative memory to stop agents from repeating specific mistakes. - Agent Configuration Linting: Integration with
agnixprovides 423 validation rules to catch configuration errors in agent/skill files before they fail in production. - Multi-Platform Support: Compatible with a wide range of AI IDEs and CLIs including Cursor, Windsurf, Roo Code, and Gemini CLI.
Related
- Project
- Project
- Project
- Project
- Project