Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
14 plugins · 43 agents · 30 skills

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. 3,750 tests. Production-grade.

Get Started View on GitHub
zsh
Paused

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. Deploying. AgentSys automates all of it.

0 Plugins
0 Agents
0 Skills
0 Tests Passing

13 Commands. One Toolkit.

Each works standalone. Together, they automate everything.

/next-task

Task to production, fully automated

  • 12-phase pipeline: discovery through deployment
  • Multi-agent review loop (code, security, perf, tests)
  • Persistent state -- resume from any phase
  • GitHub Issues, GitLab, or local task files
$ /next-task              # Start new workflow
$ /next-task --resume    # Resume interrupted workflow
How It Works

How It Works

One approval. Fully autonomous execution.

1

Pick a task

Select from GitHub Issues, GitLab, or a local task file. The agent explores your codebase and designs a plan.

2

Approve the plan

Review the implementation plan. This is the last human interaction. Everything after is automated.

3

Watch it ship

Code, review, cleanup, documentation, PR, CI, merge. All handled. You review the result.

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. 77% fewer tokens than multi-agent approaches.

One agent, one job, done well

43 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.

43 Agents. 30 Skills.

Right model for the task. Opus reasons. Sonnet validates. Haiku executes.

exploration-agentopus

Deep codebase analysis and context gathering

planning-agentopus

Step-by-step implementation design

implementation-agentopus

Autonomous code writing and modification

perf-orchestratoropus

Performance investigation coordination

perf-analyzeropus

Deep performance analysis and profiling

learn-agentopus

Web research and learning guide creation

plan-synthesizeropus

Multi-source plan synthesis and merging

agent-enhanceropus

Agent configuration quality analysis

claudemd-enhanceropus

CLAUDE.md file optimization

docs-enhanceropus

Documentation quality improvement

hooks-enhanceropus

Git hooks and automation analysis

prompt-enhanceropus

Prompt engineering best practices

skills-enhanceropus

Skill definition quality analysis

debate-orchestratoropus

Structured adversarial debate coordination

30 Skills across 14 Plugins

next-task
orchestrate-review discover-tasks validate-delivery
enhance
enhance-agents enhance-claudemd enhance-docs enhance-hooks enhance-plugins enhance-prompts enhance-skills enhance-cross-file
perf
baseline benchmark profile theory-tester code-paths theory-gatherer investigation-logger perf-analyzer
other plugins
deslop drift-analysis repo-mapping sync-docs learn-topic agnix-lint consult debate web-auth web-browse

Get Started in 30 Seconds

Recommended

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