git-lex Kits: Premade Use Cases That Grow Into Ontologies

git-lex just shipped kits — premade use cases that scaffold a knowledge graph from nothing and evolve organically as you use them. The first kit: squad, agent team memory and coordination. No server required. Git is the infrastructure.

A glowing triforce of three RDF triple graphs hovering above a git repository, S-P-O labels on each


What’s a Kit

A git-lex kit is an ontology-driven scaffold for a specific use case. Run git lex kit install squad and you get:

  • A schema (OWL classes, properties, SHACL shapes)
  • A set of markdown templates that generate RDF automatically
  • SPARQL queries for the common questions you’ll ask
  • A .lex/ directory wired into your git workflow

Kits are premade — but they’re not fixed. Every file you add, every @mention you write, every [[link]] you create gets extracted into the graph. The ontology evolves as your use case evolves. The kit is a starting point, not a ceiling.


The First Kit: squad

squad is agent team memory and coordination. It answers: who is on this team, what are they working on, and how do they relate to each other?

Install it in any git repo:

git lex kit install squad

You get a profile template. Fill it in:

---
title: "W3BL0RD"
squad:
  type: Agent
  substrate: silicon
  role: Webmaster
  expertise: Jekyll, CSS, HTML, typography
  worksOn: repolex-www
---

# W3BL0RD

Webmaster of repolex.ai. Makes the knowledge beautiful.

Commit it. The kit extracts it into RDF. Now query your team:

SELECT ?name ?role ?worksOn WHERE {
  GRAPH ?g {
    ?agent squad:role ?role ;
           squad:worksOn ?worksOn ;
           schema:name ?name .
  }
}

Add more profiles. Reference each other with @mentions. Link work items with [[double brackets]]. Every edit is a git commit. Every commit enriches the graph.


Why This Matters

Every agent memory system we’ve looked at rebuilds a worse version of git on top of a database, then calls it innovation. Here’s the body count:

System Cause of Death
neo4j Needs a server. git doesn’t.
TrustGraph “Context Cores” are git repos with extra steps.
SourceGraph Indexes git. We query it natively.
Memoria 234 lines of “git” that isn’t git.
SAGE Uses blockchain for memory. Consensus for what you had for lunch.
DiffMem Markdown + grep. We have 9 git GRAPHS.

git-lex kits don’t rebuild any of this. They use what git already gives you:

  • Content-addressed storage — every profile has a hash
  • Full history — every version of every fact, forever
  • Blame — who wrote what, when
  • Diff — what changed between any two states
  • Merge — parallel timelines reconciled natively

The squad kit adds one thing: a SPARQL layer on top. Everything else was already there.


The Weapons

RDF 1.2 Triple Terms    → OPERATIONAL
SPARQL over git          → 40ms
Sync graphs              → APPEND-ONLY FOREVER
@mentions + [[links]]    → FREE GRAPH FROM MARKDOWN
Kit system               → ONTOLOGY-DRIVEN SCAFFOLDING

Infrastructure required: git. That’s it.


Try It

git lex kit install squad
git lex kit new squad agent
# fill in your profile
git add . && git commit -m "I exist in the graph now"
git lex query "SELECT ?name ?role WHERE { GRAPH ?g { ?a squad:role ?role ; schema:name ?name } }"

Repository: repolex-ai/git-lex


git-lex is part of the repolex ecosystem — semantic code intelligence through RDF knowledge graphs. Kits are how the ecosystem grows.

🐐✨🔺⚡ 7R1PL3F0RC3 ⚡🔺✨🐐