Docs

Glossary

The team's shared vocabulary. Domain terms specific to this project - what they mean, where they came from, and a companion anti-glossary.md file for terms you deliberately don't use (and why).

The terms

Accountability Loop
The product loop detect, attribute, remember, prevent. A defect is attributed to the introducing AI session and prompt, remembered as a Lesson the agent reads from its CLAUDE.md managed block, and that Lesson prevents the next recurrence.
Aggregator
The HTTP server in the RepoOps repo (server.mjs) that spans multiple tracked repos via mirror worktrees. Specifically the RepoOps aggregator, not a generic one.
Brain
The flat collection of markdown files under .claude/brain/. Every tracked repo has one, and the aggregator has one of its own. The brain is project knowledge, not product data.
Bespoke twin
An HTML twin that is hand-curated (filters, sparklines, family trees, custom visualizations), marked with a brain-twin-bespoke meta tag so the auto-sync skips it.
Anti-glossary
The companion file anti-glossary.md for terms the team deliberately doesn't use, and why, so the shared vocabulary stays aligned on both what to say and what not to say.

These five are the set this page publishes as structured data. The full vocabulary lives in .claude/brain/glossary.md in your own repo.

See it in motion

Where to find it

  • Desktop: localhost:4000, then Memory in the sidebar, then Reference index under All tools, in the Lessons, wiki and patterns group, which lists it.
  • Hosted: repoops.ai/team/reference, from Memory in the sidebar, then Reference index under All tools, in the Lessons, wiki and patterns group, which lists it.
  • Keyboard: ⌘ K, then type “Glossary”.
  • Source: .claude/brain/glossary.md (terms), plus a companion .claude/brain/anti-glossary.md file for terms the team avoids

What it does for you

Onboarding velocity climbs.A new teammate (or a new agent) reads glossary.md and they know what "the Brain" and "an HTML twin" mean without asking.
Anti-glossary keeps the team aligned."We don't say 'self-driving' or 'autonomous coding' - those frame the product wrong" is a critical alignment tool. Both halves matter.
The AI uses your terms.CLAUDE.md links to glossary.md; the agent inherits your vocabulary instead of generic LLM defaults.

Configure

Nothing - glossary.md is a flat .md. The anti-glossary lives in its own file, .claude/brain/anti-glossary.md.

Use it well

Add a term the second time you have to explain it. Add an anti-glossary entry the second time you have to correct it. Both halves keep your AI from drifting into generic-LLM language.

Read more

Last updated