SkillFoxx

Type a name, a task or a service

AgentSys

A plugin marketplace and installer for the full dev loop: task selection, review, cleanup, PRs and releases

PluginHigh risk

agent-sh/agentsysAuthor: agent-sh

Install

/plugin marketplace add agent-sh/agentsys
/plugin install next-task@agentsys
/plugin install ship@agentsys

Other plugins install the same way by command name.

This is third-party code. Review the repository files before installing.

What it does

AgentSys ties together the agent-sh organization's plugins, each kept in its own repository, and installs them into Claude Code, OpenCode, Codex CLI, Cursor and Kiro. The /next-task command takes a task from issue selection through a worktree, plan, implementation and multi-agent review to a PR and merge. Other commands handle project review (/audit-project), cleanup of typical AI leftovers (/deslop), plan versus implementation checks (/drift-detect), performance investigation (/perf), docs sync and releases. Deterministic analysis uses regex and AST, and the model is called where judgment is needed.

Who it is for. For developers who want an agent to handle the routine around code: branches, review, CI and PRs.

Good fit when

  • You want a GitHub issue taken to a merged PR with review
  • You need to clean up agent-written code before shipping
  • You need to check the implementation matches the plan

Not a fit when

  • The project does not use git and GitHub or GitLab
  • You need one small skill rather than a whole pipeline

Example request

/next-task pick the top GitHub issue and take it to a PR

Limitations

Requires Git and Node.js 18+, an authenticated gh for GitHub workflows or glab for GitLab. Plugins are fetched from separate repositories at install time. Multi-agent pipelines use noticeably more tokens than a regular session.

How to disable. Uninstall plugins via /plugin uninstall in Claude Code or remove the agentsys marketplace; remove the global installer with npm uninstall -g agentsys.

Security check

  • Creates branches and PRs and merges in the remote repository
  • Runs shell commands and multiple agents

README in short

The README explains the approach: code does deterministic work, the model handles judgment, and findings carry certainty levels. It includes cost measurements on real tasks across models. Tables of commands and skills follow, with a detailed breakdown of each command and its agents. It ends with installation via the Claude Code marketplace or npm and requirements. MIT licensed.

FAQ

Can I use commands individually?

Yes, each command works on its own and they compose into a pipeline together.

Where is the human in /next-task?

You pick the task and approve the plan; after that the agent works on its own through merge.

Editors’ pick

A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review

PluginMedium risk286.6K
Editors’ pick

Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture

SkillLow risk262K

Reference MCP servers

Model Context Protocol servers

Editors’ pick

Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything

MCP serverMedium risk90.3K
Editors’ pick

Up-to-date, version-specific library docs and code examples in your agent's context, via MCP or a CLI plus skill

MCP serverMedium risk62K