About Wukong Code
A terminal-native, local BYOK workflow for continuing AI coding work and evaluating the current workspace with fresh evidence.
> Our Story
Give /loop a goal. It writes, checks, reviews, and fixes until the Gate returns PASS—or stops with a clear non-PASS result.
The CLI runs in the current workspace. Model requests go to the API provider you configure, under that provider's terms. Wukong does not host the model or resell its tokens.
// The current workspace wins over imported history, model confidence, or stale test claims.
> Our Mission
01Re-ground every continued task in the files and Git state that exist now.
02Keep model choice, provider cost, permissions, and runtime limits visible to the developer.
03End with complete fresh evidence, or a specific non-PASS outcome and next action.
> Core Values
[REVIEW_DRIVEN]
Deterministic checks provide facts. A fresh-context reviewer returns concrete blockers, evidence, file locations, and next actions.
[BRING_YOUR_OWN_KEYS]
You provide the model API key and choose the provider. Model requests are governed by that provider's terms; Wukong does not host the model or resell its tokens.
[TERMINAL_NATIVE]
Built for the shell, with a keyboard-driven TUI and explicit current-workspace boundaries.
[MODEL_AGNOSTIC]
Choose a supported provider and model alias for the writer and reviewer; availability follows the current provider catalog.
> How We Operate
Wukong Code is currently a free, local-first, BYOK CLI. Product claims follow the released binary, public documentation, and fresh verification evidence—not roadmap ideas or unfinished hosted infrastructure.