Terminal AI coding agent CLI> Your AI says it's done. Make it prove it._
Give /loop a goal. Wukong runs your project's own tests and checks and fixes what fails—until it passes, or tells you what's blocking.
Another agent stopped halfway? /resume picks up its work.
$curl -fsSL https://wukong.today/install.sh | shSet the goal and the limits. Wukong works toward PASS.
Three rules every Loop follows.
> DONE_MEANS_CHECKS_PASSED
PASS only when your own checks ran fresh and a separate review found nothing blocking.
> PICK_UP_WHERE_IT_STOPPED
Continue unfinished Codex, Claude Code, Cursor, Kimi, or Grok work against your repo as it is now.
> STAYS_WITHIN_YOUR_LIMITS
You set the budget. If progress stalls, Wukong stops and says why.
Build an evidence-backed coding-agent workflow
Concrete commands, safety boundaries, and evidence—not generic AI coding advice.
Run a bounded coding loop with real checks
Follow the exact install, provider, Preflight, and Loop commands for a real project.
Use your own model API key
Understand local orchestration, provider billing, privacy boundaries, and Loop limits.
Hand off work from five supported coding agents
See what is imported, what is never inherited, and how Wukong re-checks the live workspace.
▄▄█████▄▄
▄█████████████▄
▄▄█████████████████▄▄
▄███████████████████████▄
▀███████████████████████▀
▀▀█████████████████▀▀
▀███████████▀
▄▀▀▄ ▄███████████▄ ▄█▀
█ ▄██▀███████▀██▄██
▀▄▄▄▀██ ▀█▀█▀ ████▀
▀▀▀ ▀▀▀ ██