Home Scan Diff Docs

Scan a project

Point at a local folder or a public GitHub URL. The scan produces a self-contained HTML report you can keep, share, or open offline.

Scan a folder from your computer

Pick a folder on this machine — your browser reads it and uploads the source to the full server-side scanner. Use this when the server can't see your disk (e.g. the hosted site can't reach G:/… on your PC). .env, node_modules, .git, build dirs and binaries are skipped automatically. Capped at 5000 files / 20 MB.

Tip: running diagram3 locally (scan.bat, or node index.js "G:/temp/sim3_prod") scans a local folder directly — no upload needed.

— OR — point the server at a path it can already reach —

Local folder path (a path on the server, not your PC)

Absolute path. Leave empty if scanning a GitHub URL instead.

GitHub URL (optional)

Shallow-cloned to the cache. Re-scans pull latest.

Exclude patterns (comma-separated globs)

Core

AST mode (accurate JS/TS/Py) Git metadata per file Embed source for AI chat

Output

Split JSON export SARIF SBOM (CycloneDX)

Enrichers Enable all missing binaries are silently skipped

opengrep (security rules) CVEs via OSV.dev (built-in) osv-scanner (lockfile CVEs) gitleaks (secrets) knip (dead code) skott (unused deps) oxlint (520+ lint rules) scc (LOC + complexity) fta (maintainability) bearer (privacy / PII) ts-morph (type-aware edges)

running

Stuck on "A scan is already running"? Click force-release lock — kills any running child and resets the server state.

✓ Scan complete

View raw HTML ↗

Scanner output

DiagraM3 · local-first codebase intelligence · MIT