TAUSIK

介紹

AI coding agents can't quietly fake "done". Signed ed25519 receipts prove the gates actually ran — bound to the commit, verifiable offline. Fail-closed hooks block the edit, not the PR. Harness engineering over Claude Code, Cursor, Codex, Qwen, OpenCode.

翻譯準備中

取自儲存庫 README

README

正在載入完整 README。 以下為已儲存的摘錄。

**AI coding agents that can't quietly fake "done."** TAUSIK sits on top of the AI coding agent you already use — Claude Code, Cursor, Codex and others — and checks its work. The agent must open a task with a definition of "done" before it edits code, and must show passing checks before it closes the task. Every decision and every failed approach is remembered for the next session. For developers who don't want to take "tests pass" on faith.