Repolode
Submit
agent-sh

agent-sh/agentsys

A modular orchestration system of plugins, agents and skills that automates task selection, code review, PR shipping and docs sync for AI coding agents.

View on GitHubWebsite
Stars
1,004
Forks
118
Open issues
0
Last commit
15 hours ago

What is it?

AgentSys is a runtime and marketplace for AI coding agents. It bundles plugins that handle the work around writing code: picking tasks from GitHub Issues or other sources, creating isolated worktrees, running multi-agent code review, cleaning up AI-generated artifacts, monitoring CI, addressing PR comments, and merging. Each command runs standalone or chains into end-to-end pipelines with phase gates and state that persists across sessions.

The system targets Claude Code, OpenCode, Codex CLI, Cursor, and Kiro. It is installed through a marketplace or an npm installer, which fetches each plugin from its own repository under the agent-sh organization. Findings are graded by certainty (HIGH, MEDIUM, LOW), so auto-fixes are limited to high-confidence issues.

Beyond the pipeline commands, the project includes tools for analyzing repository history and symbols, linting agent configuration files through the agnix integration, performance investigations with baselines and profiling, cross-tool consultation and debate between AI CLIs, and release automation across many package ecosystems.

Key features

  • 01Orchestrates task discovery, implementation, review, and PR merge as gated pipelines.
  • 02Runs parallel multi-agent code reviews with severity-graded findings.
  • 03Detects and removes AI slop such as debug statements and placeholder comments.
  • 04Monitors CI, fixes failures, and replies to PR review comments.
  • 05Lints agent configuration files such as SKILL.md, CLAUDE.md, and hooks via agnix.
  • 06Works across Claude Code, OpenCode, Codex CLI, Cursor, and Kiro.

Use cases

  • Automating the path from a GitHub issue to a merged pull request with AI agents.
  • Running pre-ship quality gates such as review, test coverage checks, and docs sync before opening a PR.
  • Maintaining and linting custom skills, agents, and prompts for AI coding tools.
  • Consulting or debating between different AI CLI tools on a design decision.

Topics

#ai-agents#claude-code#automation#code-review#developer-workflow#plugins#cli#agent#ai#anthropic

FAQ

What is agentsys?
A modular orchestration system of plugins, agents and skills that automates task selection, code review, PR shipping and docs sync for AI coding agents. AgentSys is a runtime and marketplace for AI coding agents. It bundles plugins that handle the work around writing code: picking tasks from GitHub Issues or other sources, creating isolated worktrees, running multi-agent code review, cleaning up AI-generated artifacts, monitoring CI, addressing PR comments, and merging. Each command runs standalone or chains into end-to-end pipelines with phase gates and state that persists across sessions.
Is agentsys free to use? What license does it have?
agentsys is open source under the MIT license. Check the license terms before using it in commercial products.
Is agentsys still maintained?
Actively maintained. The last commit to agentsys was on Oct 18, 2026.
What can I use agentsys for?
Automating the path from a GitHub issue to a merged pull request with AI agents. Running pre-ship quality gates such as review, test coverage checks, and docs sync before opening a PR. Maintaining and linting custom skills, agents, and prompts for AI coding tools. Consulting or debating between different AI CLI tools on a design decision.
Where can I get agentsys?
The source code and installation instructions are on GitHub: https://github.com/agent-sh/agentsys

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects