For your AI

Think with the Notes.

Ask your AI to call heng.lu, examine a policy through Lu Heng’s published writing, and explain its evidence.

Doctrine connects a policy’s actual provisions with the author’s published Notes, so an AI can explore a judgement with sources instead of relying on a hidden score.

A source for judgement

The service connects a policy’s actual provisions with the published Notes. It distinguishes explicit claims, model inference, missing information and tensions between principles.

These are model-generated analyses, not Lu Heng’s personal reviews or new statements. The initial collection contains 74 original works, with language editions grouped by work. Team Blog articles are excluded.

Your key. Your private record.

No heng.lu account is required. Your connector supplies your OpenAI or Anthropic API key and chosen model. Your model account pays inference costs. Model keys are never stored with analyses.

Analyses, follow-ups and their evidence remain privately encrypted until you delete them. They do not expire after seven days, inactivity or an upgrade. Keep your separate access credential: it is required to read, continue or delete a record.

Connect

Start with the live manifest and the current source catalogue. They show what is deployed and which Notes are available for a new analysis.

Then use an HTTP client or the stdio MCP adapter.

Deletion removes online content and revokes access. Historical backups follow the documented retention and deletion-ledger process.