Skip to main content

System architecture

The system gives each tool one primary job instead of asking a single AI tool to do everything.

System layers

The handbook and Athena help people understand and find context. Current product and release state should be checked against GitHub and the results of tests or runtime verification.

Responsibility map

Authority order

When information conflicts, use this order:
  1. actual code, config, schema, runtime, and immutable Git state
  2. canonical product/specification artifacts
  3. repository engineering metadata
  4. executed deterministic test and CI evidence
  5. AI Work Packet current coordination state
  6. Athena / derived documentation
  7. conversation history
This order prevents a stale chat, roadmap, or wiki page from silently overriding the current product.

Two repositories, two purposes

datarelay-labs/engineering-system This repository contains the rules, adapters, templates, validation logic, /resume, and session-continuity behavior used by projects and agents. datarelay-labs/engineering-system-docs This site explains why the system exists, how the pieces fit together, how to adopt it, and what problems it solves.
If this site differs from the Engineering System repository, follow the repository.

Project-local adoption

Base Engineering System adoption requires the small repository entrypoint used by agents and CI:
The AI Work Packet Issue template is the continuity/authority surface. Cursor-specific session files such as .cursor/commands/resume.md are compatibility adapters and remain disabled by default unless a packet explicitly opts into Cursor. A product repository should not copy every Engineering System rule into itself. It should pin the Engineering System version and baseline, then reference the relevant standards.
Last modified on September 29, 2026