A modular runtime and orchestration system
for AI agents.
Structured pipelines, gated phases, specialized agents. Works with Claude Code, OpenCode, Codex CLI, Cursor, and Kiro. Tested on Linux and Windows. Production-grade.
AI models write code.
That's not the hard part anymore.
The hard part is everything else. Picking what to work on. Managing branches. Reviewing output. Cleaning up AI artifacts. Handling CI. Addressing reviewer comments. AgentSys automates all of it.
8 Commands. One Toolkit.
Each works standalone. Together, they automate everything.
/foreman
GitHub issues as a task queue
- Picks the next ready issue and claims it
- One worktree subagent per issue, within host capacity
- Reads Revuto's findings on every PR
- Lands at the exact head it checked, through the repo's land command
$ /foreman # Run the queue until nothing is ready
$ /foreman status # Counts, claims and capacity
$ /foreman 123 # Work one issue
/deslop
Kill AI slop before it ships
- 3-phase detection: regex, multi-pass analyzers, CLI tools
- Certainty-graded findings (HIGH / MEDIUM / LOW)
- JS/TS, Python, Rust, Go, Java
- Auto-fix HIGH certainty issues
$ /deslop # Report only (safe)
$ /deslop apply # Fix HIGH certainty issues
/agnix
Lint agent configs before they break
- 757 validation rules across 46 categories
- 10+ AI tools: Claude Code, Cursor, Copilot, Codex, OpenCode, Gemini CLI
- 125 auto-fixable rules with --fix flag
- SARIF output for GitHub Code Scanning
$ /agnix # Validate current project
$ /agnix --fix # Auto-fix fixable issues
/learn
Research any topic, build a learning guide
- Progressive discovery: broad to specific to deep
- Quality-scored sources (authority, recency, depth)
- Structured guide with examples and best practices
- RAG index for future agent lookups
$ /learn react hooks --depth=deep # Comprehensive
$ /learn kubernetes --depth=brief # Quick overview
/skillers
Learn from your workflow patterns
- Reads transcripts from Claude Code, Codex, OpenCode
- Clusters patterns into themed knowledge
- Suggests skills, hooks, and agents to automate repetitive work
- No per-turn overhead - works from saved transcripts
$ /skillers # Analyze workflow patterns
$ /skillers compact # Compact transcripts into knowledge
/banthis
Durable negative memory
- Persists banned agent behaviors into CLAUDE.md or AGENTS.md
- Maintains a managed section with add, list, show, remove
- Includes init meta-rule so agents invoke it automatically
- Works project-local or global with explicit flags
$ /banthis "stop ending with vague optional offers"
$ banthis list
/skill-curator
Design reliable SKILL.md files
- Writes strong trigger descriptions and tight scopes
- Uses router patterns for broad skills
- Adds concrete Skip unless gates for pattern rules
- Keeps skills cross-tool and agnix-ready
$ /skill-curator "create a skill for reviewing background jobs"
$ /skill-curator --improve path/to/SKILL.md
/system-prompt-curator
Design autonomous agent prompts
- Matches identity to task and tool surface
- Builds phased workflows with evidence gates
- Includes realistic error recovery examples
- Separates prompt guidance from harness checks
$ /system-prompt-curator "GitHub issue resolver"
$ /system-prompt-curator --improve prompt.md --for-orchestrator
Built Different
Not another AI wrapper. Engineering-grade workflow automation.
Code does code work. AI does AI work.
Static analysis, regex, and AST for detection. LLMs only for synthesis and judgment.
One agent, one job, done well
5 specialized agents, each with a narrow scope and clear success criteria. No agent tries to do everything.
Pipeline with gates
Each step must pass before the next begins. Can't push before review. Can't merge before CI. Hooks enforce it.
Validate plan and results
Approve the plan. See the results. The middle is automated. One approval unlocks autonomous execution.
5 Agents. 14 Skills.
Right model for the task. Opus reasons. Sonnet validates.
Workflow pattern analysis and automation suggestions
AI slop pattern detection and cleanup
Web research and learning guide creation
Agent config linting orchestration
Transcript compaction into knowledge themes
14 Skills across 11 Plugins
Get Started in 30 Seconds
Recommended
$ /plugin marketplace add agent-sh/agentsys
$ /plugin install foreman@agentsys
Interactive installer for Claude Code, OpenCode, and Codex CLI
$ npm install -g agentsys && agentsys
Clone and install from source
$ git clone https://github.com/agent-sh/agentsys.git
$ cd agentsys
$ npm install