| English | 简体中文 |
Public mission, essays, diagrams, evidence indexes, and whitepaper drafts for a governed persistent multi-agent runtime.
面向受治理的持久化多智能体运行时的公开世界观、文章、图表、证据索引与白皮书草稿仓库。
| Language | Entry |
|---|---|
| English | README.en.md |
| 简体中文 | README.zh-CN.md |
This is the public documentation and publication home for constitutional_runtime.
It argues that long-running AI work cannot become reliable only by giving one agent more memory, more tools, or a larger context window. It needs a governed runtime where mission ownership, authority relations, evidence gates, audit receipts, state transitions, and lawful closure are explicit runtime objects.
本仓库不是实现源码仓库。它公开的是概念、文章、白皮书路线、图表、引用证据与发布包;私有源码、私有 prompt、内部日志、凭证和运行痕迹不在这里公开。
The repository now includes active GitHub-facing configuration in .github/:
Authoring and policy copies remain under .internal/ so release checks can keep the public boundary explicit.
For Chapter 1, each language edition’s PDF is the final authority for both text and layout. If Markdown, GitHub Pages, Substack, Hugging Face, or Wiki editions diverge from the matching PDF, correct the platform edition against the PDF.
For Chapter 2 v7.3, the supplied source is preserved as chapter2.html, with a repository-readable chapter2.md conversion and extracted figure assets. If a canonical PDF / DOCX package is generated later, update the Chapter 2 version map to name that artifact as the layout authority.
Allowed here:
Not allowed here:
Run the public readiness check before publishing:
make -f .internal/engineering/Makefile check