ontology-atlas

Desktop app plus MCP server keeping a codebase's ontology as reviewable Markdown in the repo: domains and capabilities linked to code evidence, compiled to a graph for blast-radius queries.

85 14 TypeScript MITupdated today

knowledge-graphs code-intel coding

View on GitHubDispute this mapping →

Curator's take

Pick it when the expensive question is *why* the code is shaped this way and what a change will touch. Because the ontology is authored and reviewed rather than inferred, unknown coverage shows up as unknown instead of a confident wrong graph — and every agent write lands as a diffable line in a Markdown file. It is not a code indexer: it will not find symbols or callers, and it only knows what someone wrote down, so an unmaintained atlas/ rots quietly while still looking authoritative. Practical caveat: this is a desktop install with an unsigned Windows beta; Linux and CI users run the CLI and MCP server from source.

Mapped by ShipWithAI editors · links verified

README.md

Ontology Atlas

Understand what your codebase builds, why it is structured that way,
and what a change will affect.

Download for macOS · Windows x64 beta

_{Windows is a public, unsigned beta. Microsoft Defender SmartScreen may
warn about an unknown publisher, and managed work PCs may block it.}

The current Ontology Atlas macOS app with the Online Store project selected: the domains it contains named around it, everything unrelated receding, and the right inspector showing the project record, its code-evidence state, and the offer to connect a code folder

_{The installed macOS app reading the example vault in
samples/storefront — an online
store, written as nothing but Markdown files in a folder. The live demo and
feature inventory are the current behavior
contract.}

Ontology Atlas keeps that explanation as a codebase ontology in repository Markdown: product domains and capabilities linked to implementation evidence, dependencies, and impact. People judge the files and git diffs; AI agents query and maintain the same ontology over MCP.

Each desktop download includes the Atlas app and its MCP server. One button writes the connection files; the restart and mcp-verify steps then prove the live connection from your agent's working folder.

Live demo · Guide · The journey · Where it stands · What this is not · Status

In 30 seconds

Source code shows how a system works. It rarely preserves which product capability the code serves, why its boundaries exist, or what a change could affect. Atlas keeps those answers beside the code, in an atlas/ folder of Markdown files inside the repository itself — so the meaning is cloned, branched, reviewed, and shared with the code rather than alongside it.

Use Atlas before a change so a person and an AI agent start from the same answer: what the code is for, where to begin, what else it touches, and what to verify. That answer stays bounded — a list of observed capabilities is not treated as exhaustive, and unsupported scope remains visible uncertainty.

Each file's frontmatter declares what it is (project, domain, capability, element, or a linked document) and what it points at. That folder is the whole database.

Because the kinds and relation types are a small fixed set, the folder is not just readable — it is computable. Atlas compiles it into a graph and answers questions a note-taking tool cannot: *what breaks if I change this, what is this capability's

Continue your stack

What teams reach for next — and why each earns a place beside ontology-atlas. Ranked by curator confidence.

Pairs well with

serena

The IDE for your coding agent: an MCP toolkit giving symbol-level retrieval, editing, refactoring and debugging over real language servers — or a JetBrains plugin backend.

Why it fitsSerena gives an agent symbol-level retrieval and editing through real language servers; Atlas supplies the product-level layer above symbols — which capability this code serves and what else depends on it. Plan in Atlas, execute in Serena.

vault-ld

An open spec for Markdown vaults as linked data: YAML-LD frontmatter plus a shared @context makes notes an RDF graph, with reference scripts that round-trip vault to Turtle and back.

Why it fitsAtlas's atlas/ folder is exactly the pattern vault-ld specs — a Markdown vault whose frontmatter is a graph. Adopt vault-ld's YAML-LD context if you want the ontology to round-trip to RDF and join other linked data.

Alternative to See all + compare →

Graft

Context layer for large codebases: a graph of plain-English markdown nodes — no embeddings, no index — agents read like any repo file. Claude Code hooks + MCP; 42% fewer tokens in its bench.

Why it fitsBoth hand agents a hand-written Markdown context layer over a large codebase instead of embeddings. Graft is a flat plain-English node graph with Claude Code hooks; Atlas constrains you to five node kinds and a fixed relation set so the folder becomes computable.

codegraph-mcp

No-train, on-prem code knowledge graph served to AI agents over MCP — symbols, call edges, cross-language links and blast-radius queries, with a hash-chained audit log of every read.

Why it fitsBoth answer blast-radius questions over MCP. CodeGraph derives the graph automatically by parsing code, so it is always current but only knows structure; Atlas curates it by hand, so it captures intent but can drift from the code it describes.

codeflow

Paste a GitHub URL or drop a local folder → interactive architecture map in the browser: dependency graph, blast radius, health grade, security scan. Single index.html, zero install.

Why it fitsBoth produce a navigable architecture map of a repo. Codeflow derives one in the browser from a URL in minutes with zero upkeep; Atlas makes you write and review it, trading effort for an explanation the parser can't infer.