How Aura Knowledge is operated.

Publication contract

The essay is public only after the record can support it.

The site treats an article as more than prose. A publishable article has a human page, an agent brief, a JSON file, source notes, source lists, review history, and a content hash tying approval to the exact text that went live.

• Humans get a calm reading path before seeing machine-oriented files.
• Reviewers can check important points, sources, and maturity state.
• Agents get stable files without taking over the public navigation.

Public surfaces

Reader pages and technical pages are separated on purpose.

Repositories

The public source explains how the public knowledge is made.

The current repository, aura-knowledge.github.io, contains the site, content sources, schemas, generated discovery files, and publication scripts. Future repositories can extend the same idea with focused tools or experiments.

• content/articles keeps the human article, machine brief, and article record together.
• public/agents exposes generated machine-readable files.
• public/graph exposes topic, article, and source-backed point relationships.

Principles

The operating model stays human-first and evidence-aware.

• Readable pages are the front door.
• Important points should be checkable without turning the page into a database UI.
• Sources and review history are part of the record, not an afterthought.
• Maturity labels should be honest about whether an idea is early or stable.
• Machine-readable files should support agents without becoming the human path.

Agents

Automation has a separate entrance.

Agent-facing files exist so tools can retrieve compact metadata, point IDs, source IDs, article files, roadmap files, and map data without scraping the human pages first.

Planning

The roadmap is a build note, not a reader tab.

The roadmap remains public because it explains future product and research work. It is intentionally placed here and in the footer instead of the primary navigation.