agentsys
Orchestrate AI agents through full coding workflows from task to merge
All Skills in This Repository (31)
Pure Emerald Level Indicatorsmaintain-cross-platform
Coordinate cross-platform release checks and configuration alignment across Claude Code, OpenCoder, and Codex.
perf-benchmarker
Run sequential performance benchmarks with strict duration and warmup rules.
deslop
Detect and remove AI-generated artifacts from codebases with structured JSON output.
perf-theory-tester
Run a controlled performance test comparing metrics against baseline data.
enhance-hooks
Automate safety and correctness reviews for Claude Code hook definitions and scripts.
perf-analyzer
Synthesize performance investigation results into structured, evidence-backed recommendations.
enhance-claude-memory
Analyze CLAUDE.md and AGENTS.md files to identify redundancy and propose memory-optimization edits.
enhance-skills
Analyze SKILL.md files and report quality issues with optional auto-fixes.
enhance-cross-file
Detect cross-file inconsistencies across prompts, frontmatter, agents, and workflows.
learn
Generate structured learning guides with ranked online sources and study sections.
debate
Orchestrate structured multi-round debates between AI tools with evidence-backed verdicts.
perf-baseline-manager
Store a single performance baseline JSON per version in a state directory.
Frequently Asked Questions
FAQPage SchemaHow to install AgentSys?▼
Run `npx skills add agent-sh/agentsys --all -g -y` in your terminal to install the full suite globally for your AI coding tool.
What does AgentSys do?▼
It orchestrates specialized AI agents through end-to-end development workflows, from picking a task to writing code, reviewing it, and merging the PR. Phase gates and persistent state keep work on track across sessions.
Which AI coding tools work with AgentSys?▼
AgentSys runs on Claude Code, OpenCode, Codex CLI, Cursor, and Kiro. The installer automatically adapts commands, agents, and skills to each platform.
Can AgentSys clean up AI-generated code?▼
Yes. The deslop and enhance plugins detect debug statements, dead code, and prompt quality issues with certainty levels, then auto-fix safe findings.
Do I need coding experience to use AgentSys?▼
Basic Git and terminal familiarity helps, but the agents handle task selection, implementation, and PR creation from plain-English requests.
Related Repositories in Software Engineering
View All in Software Engineering→openclaw
Run a personal AI assistant across your devices and chat apps
superpowers
Gives coding agents a disciplined workflow from idea to merged code
react
AI agent skills for building, testing, and porting React core