The side panel is the product.

Open Magellan next to Setup and it analyzes whatever you click. Or search any component by name and pull it up without leaving the panel.

40+ RULES

The PMD catalog, in the browser

Apex: SOQL and DML in loops, async jobs in loops, missing with sharing, CRUD/FLS gaps, SOQL injection, hardcoded IDs and credentials, insecure endpoints, empty catches, tests with no assertions, SeeAllData=true, cyclomatic complexity, deep nesting, long methods. Flow: DML and Get Records inside loops, missing fault paths, System-Mode-Without-Sharing, hardcoded IDs and URLs. LWC: innerHTML, global document queries, unhandled GraphQL errors, leftover console.log.

A REAL TOKENIZER

Not grep. A lexer and a structural pass.

Magellan tokenizes Apex — comments, strings, and inline SOQL are tagged, so a query quoted in a comment never trips a rule — then builds a loop and method model to ask structural questions: is this query inside any loop? Every finding carries a confidence; heuristics that would need a full parser are shipped honestly labelled heuristic, never presented as certain.

RANKED, EXPLAINED, FIXABLE

Line, why, and how to fix it

Findings sort critical → warning → info, then by confidence, then by line. Each one shows the offending snippet, a plain-English explanation of the risk, and a concrete fix hint. Click through to that exact component in Setup. Search any class, trigger, flow or LWC by name from the panel itself.

Atlas: the whole org, ranked.

One button scans every Apex class, trigger, active Flow, and LWC in the org and hands you the tech-debt map a lead or an auditor actually wants.

Magellan Atlas: org-wide dashboard with severity counts, type and severity filters, a ranked cross-org findings list, a What's on fire card, worst offenders and top rules fired — Every finding across the org, filterable — with worst offenders, top rules fired, and what's on fire right now.

ORG SCAN

Same rules, whole org

Atlas runs the exact same engine over the org inventory — no new logic, no server. Filter by type or severity, search findings, sort by rule. Worst offenders rank components by severity-weighted debt; top rules fired tells you which anti-pattern dominates. Export CSV or JSON for the ticket.

WHAT'S ON FIRE

Risk vs. burning

Static findings say a class could fail. Atlas also reads your recent failed and aborted Apex jobs — batch, queueable, scheduled — and shows what's actually failing in production. A risky-but-working class and a class that aborted 169 times last week are very different problems; Magellan shows you both, and flags the component that's on both lists first.

WHICH ORG?

Never scan the wrong org

Signed into several orgs? Every surface names the org it's reading, with a loud PRODUCTION or Sandbox badge, and Atlas has an org switcher. The scan screen itself confirms the target while it runs.

Explore your org as a world.

Magellan circumnavigated the globe. So does this: your org's code health as a 3D world map you spin, zoom, and fly through.

Magellan Universe: a 3D globe where continents are component types coloured by code health, hexagon tiles are components, flames mark failing jobs, and two satellites named Marco and Polo orbit — Continents are your Apex, Triggers, Flows and LWC — coloured by weighted health. Flames are jobs failing right now.

Text first, 3D second: the globe is how you find the problem; the side panel is how you read it.

Continents are component types, sized by how much code lives there and coloured by weighted overall health — lush green through arid tan to scorched rust. One critical class can't paint five hundred clean ones red.
Hex tiles are components, clustered into severity districts — the critical city, the warning city — with the worst offenders at each city's centre. Tile size is the finding count.
Managed packages are an archipelago: one island per installed package, neutral gray because vendor code isn't linted — until something in it burns.
Trade routes are dependencies. Focus a component and its routes light up: cyan for what it uses, ember for what depends on it. Load-bearing debt, visible.
Flames are live failures. The terrain says could burn; the fire says is burning — hotter and faster the more recent and frequent.
Voyage mode sails port to port through your worst offenders. Marco and Polo keep orbit overhead.

The Magellan globe from orbit — Search any component and the camera pulls back, spins the planet, and dives onto its tile.

No pipeline. No login. No config.

Every other analyzer wants a checkout, a scan job, or an upload. Magellan runs on the session you already have, against the code already in your browser.

CLICK, DON'T CONFIGURE

Zero setup

Nothing to install in the org — no connected app, no OAuth, no package, no CI step. Sign into Salesforce in Chrome the way you already do. Open the panel. Click.

RIGHT NOW, NOT NEXT BUILD

Feedback while you're looking at it

PMD and Code Analyzer tell you in a pipeline, later. Magellan tells you in Setup, now — the moment you open the class, before it's committed anywhere.

READ-ONLY, LOCAL

Look, don't touch

Magellan only ever reads. Source and metadata come from your browser straight to your org and stay there — analysis runs client-side, no servers, no accounts, no telemetry.

Apex classesApex triggersFlowsLWC Failed & aborted jobsDependenciesManaged packages

Read-only isn't a setting. It's the product.

Magellan never writes to an org, and your code never leaves your browser.

The rules run in the browser. The engine is pure client-side code — it can't make a network call even if it wanted to. The org-wide scan is exactly the moment other tools upload your whole codebase to a cloud; Magellan doesn't.
No servers, no accounts. The only calls are from your browser to your own org's APIs, using the session you already have.
No telemetry. We don't know when you install it, open it, or what orgs you point it at.
Nothing installed in your org. No connected app, no OAuth, no package.

Read the full privacy policy →

Magellan PMD-grade code review, the moment you click a component.