Free and Apache 2.0. Runs on your machine; no model, no embeddings, nothing uploaded.View on GitHub

Cross-repository code graph

Every repository, language and framework. And how they connect.

Enola is an open-source CLI and MCP server that reads your source code and links what it finds, inside each repository and across them. One local binary. No model, and nothing leaves your machine.

curl -fsSL https://raw.githubusercontent.com/enola-labs/enola/main/install.sh | sh

Then, from inside any repository: enola --explain .
Read-only. No config, no account, nothing written to disk.

enola --explain
on enola's own repository
$ cd enola
$ enola --explain .

Overview
  Languages:           go, typescript, c, ruby, python
  Total facts:         10814

Architecture
  cyclic dependencies         0
  layer violations            0

Impact analysis (hotspots)
  Top hotspots (by coupling):
    module            fan-in  fan-out crit  blast radius
    internal/facts       253        1 high  96
    pkg/command            1       94 high  1
    pkg/bootstrap         15       69 high  4

Across repositories

Linked across repositories. Even when the paths don't match.

Give Enola a backend and the apps that call it, and each call is linked to the code that serves it. Change an endpoint and you see who calls it, in every repository.

web repository · client.go
func fetchOrder() {
    http.Get(apiBase + "/api/v2/orders/{id}")
}
api repository · server.go
func main() {
    v2 := r.PathPrefix("/api/v2").Subrouter()
    registerOrders(v2)
}

func registerOrders(r *mux.Router) {
    r.HandleFunc("/orders/{id}", getOrder)
}

The path the web app calls is never written in the api repository. It is split across two functions, and Enola puts it back together. A URL built at runtime is reported as unresolved, not guessed.

How repositories are linked, and where linking stops →

Four ways to use it

One graph. Use it wherever you work.

On your own

See what a pull brought in. No agent, no AI.

enola baseline pin
git pull
enola check

In CI

The same check on every pull request, annotated on the lines that caused it.

uses: enola-labs/enola-action@v2 enola-action →

With your coding agent

The agent asks what depends on the code before it edits, and checks itself after.

enola install --hooks
claude mcp add enola enola
All MCP clients →

As a foundation

A snapshot is plain files in a documented format. Cognee builds its code search on it.

Building on Enola →

Checking a change

Tests check behavior. Enola checks structure.

Pin the graph, make the change, compare. Enola reports only what the change introduced, and fails only the rules you chose.

enola check
compiles, tests pass, layer order broken
$ enola check --fail-on=layers

FAIL — 1 structural regression introduced.

Regressions (fail):
  - [layers] 1.00 — Layer violation: storage -> delivery
      import of notify

$ echo $?
1
239findings already in a benchmark repository
→
1reported: the one this change introduced

Existing debt stays silent, so the check can stay on.

Baselines, exit codes and what can fail →

Measured, not claimed

Public benchmarks. Scripts included.

23languages and formats in one graph. Discourse's Ruby and Ember halves, linked.
95/95public repositories produced byte-identical graphs across a cold run and two warm ones
20/20injected cycles caught, with none of 1,859 pre-existing findings repeated as new

Free, local, open source

The whole engine. No license check.

Apache 2.0. Every language, tool and check ships in the open-source repository.

Running Enola across many teams and repositories?

Roll out Enola across your organization →