Orchestrate a Plan-Execute-Review development loop where a brainstorming agent, executor agent, and adversarial code reviewer collaborate to build and refine workflows from natural-language descriptions, with cloud sync, session pause/resume, and OAuth authentication.
Cancel the active dev workflow loop (archives by default, --hard to wipe)
Resume a paused dev workflow, or take over a cloud session from another machine
Create a new workflow suite (published to hub by default), or edit an existing one with --flow=<path>
Pause the active dev workflow loop without clearing state — resume later with /stagent:continue
Sign in to the stagent cloud server
Create a new workflow suite from a natural-language description, or edit an existing one when --flow=<path> is passed. Dispatches the create-workflow stagent (plan → write → validate loop) — does not write files directly.
Bootstrap a dev workflow session (for /stagent:start): parse flags, derive a topic from the task description, call setup-workflow.sh, and hand off to the workflow loop skill. Does NOT drive the state machine — that's stagent:stagent's job.
Drive the dev workflow state machine: read state.md, execute the current stage (inline or subagent), transition via update-status.sh, loop until terminal. Precondition: state.md already exists (some upstream caller bootstrapped it). Does NOT bootstrap.
Matches all tools
Hooks run on every tool call, not just specific ones
Executes bash commands
Hook triggers when Bash tool is used
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Modifies files
Hook triggers on file write and edit operations
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Uses power tools
Uses Bash, Write, or Edit tools
English | 简体中文 | 日本語 | 한국어 | Français | Deutsch | Español
A Claude Code plugin that runs config-driven development workflows as a state machine. You declare stages, transitions, and inputs in a single workflow.json; the plugin's hooks and scripts drive the loop.
Two modes:
<project>/.stagent/, no network.Run this command
claude plugin marketplace add jie-worldstatelabs/stagent && claude plugin install stagent@stagent
Already installed? Update with:
claude plugin marketplace update stagent && claude plugin update stagent@stagent
Requires: Claude Code, jq, curl, git (cloud mode also relies on standard POSIX tools like sha256sum / shasum).
Optional but recommended: sign in first to claim session ownership and better manage your past sessions.
/stagent:login
Start the default development workflow — it builds what you describe:
/stagent:start --flow=cloud://demo "Build a journaling app with MBTI insights inferred from journal entries"
The skill prints a live UI URL. Without sign-in, this is an anonymous, publicly viewable session — anyone with the link can watch the state machine run in real time (stage timeline, rendered artifacts, git diff baseline..HEAD updating via SSE), and there is no owner.
For a fully offline run, switch to local mode:
/stagent:start --mode=local "Build a journaling app with MBTI insights inferred from journal entries"
Define your own workflow from a natural-language prompt — stagent scaffolds the stages:
/stagent:create "plan, implement, critique & score UX"
This defaults to cloud mode: the new template is published to your hub account after the planning + writing stages finish. Sign in first if you haven't already:
/stagent:login
For a fully offline run (template stays on disk under ~/.config/stagent/workflows/<name>/, nothing pushed to the hub), switch to local mode:
/stagent:create --mode=local "plan, implement, critique & score UX"
Need inspiration? Browse the cookbook for twelve battle-tested workflow templates you can fork or remix.
With no --flow flag:
cloud://demo from the hub — a hosted template that may evolve independently of this READMEskills/stagent/workflow/ (offline fallback) — the canonical source for the cycle described belowThe bundled workflow runs a plan → execute → review → QA → deploy cycle:
baseline git SHA captured at workflow start. PASS → QA; FAIL → loop back to Execute.vercel whoami, vercel link on first run, sync production env vars, vercel --prod, smoke-check the URL. Interruptible because first-run setup may need vercel login in another terminal or env-var values from you. Done → terminal complete.The execute → review → QA loop runs autonomously after you approve the plan. A Stop hook guarantees the loop runs to completion (until QA passes; deploy then runs as the final, interruptible stage). The loop stops on one of: deploy completes (terminal complete), max_epoch is hit (default 20, configured in workflow.json → .max_epoch; breaks runaway iteration by forcing terminal escalated), or you intervene with /stagent:interrupt (pauses) or /stagent:cancel (terminal cancelled). All three — complete, escalated, cancelled — are declared in workflow.json → .terminal_stages.
npx claudepluginhub jie-worldstatelabs/stagent --plugin stagentStructured project planning and execution through brainstorm, spec, and build phases across three execution tiers: sequential, delegated sub-agents, and full agent teams
Describe your goal, approve the spec, then step away — Claude and Codex loop together until it's right.
Plan-Build-Run — Structured development workflow for Claude Code. Solves context rot through disciplined subagent delegation, structured planning, atomic execution, and goal-backward verification.
Production-ready Claude Code configuration with role-based workflows (PM→Lead→Designer→Dev→QA), safety hooks, 44 commands, 19 skills, 8 agents, 43 rules, 30 hook scripts across 19 events, auto-learning pipeline, hook profiles, and multi-language coding standards
Hooks-based workflow orchestration for solo product development with Claude Code
Agentic engineering done right — 57 structured workflows, 17 specialist agent personas, persistent memory across sessions, integrated learning partner, and impeccable UI design system. Works with Claude Code, Windsurf, Cursor, Gemini CLI, OpenCode, and Codex.