Understanding an unfamiliar codebase means juggling five tools with five output formats, and some of them want your source uploaded.

1. 01

   An interface built for exploration

   A survey-map visual identity, a repository switcher and grouped navigation connect dedicated views for overview, activity, commits, contributors, branches, files, search, metrics, dependencies and duplicates. Light, dark and system themes, keyboard-accessible repository picking and a mobile menu carry the same interface across screens.

2. 02

   History with the right clock

   Commit calendars, hour-of-week activity and streaks use each author's local time. A paginated commit log links to contributors and file ownership, while a lazy file tree and sortable table expose per-file size, complexity and history.

3. 03

   Findings you can trace

   Architecture graphs collapse imports to folders at a chosen depth, lay out importers beside their dependencies and let you select a node to trace its connections. Cycle and potentially dead-file findings, duplicate code samples with linked locations, and highlighted search matches lead back to the source. Graphs export as SVG.

4. 04

   A scan you can watch and control

   The Go pipeline streams and batch-inserts files, history, dependencies, imports and duplicate blocks, then builds a SQLite FTS5 content index. Persisted background jobs report progress over WebSocket, support pause, resume and cancel, and re-queue interrupted scans after a crash. Completing a scan refreshes the open views.

5. 05

   Never leaves the machine

   No account, external APIs, telemetry or source uploads. The API binds to loopback by default, the frontend talks to localhost, and WebSocket connections are limited to the same origin. Files, commits and contributors export as CSV or JSON.

3,602

files scanned

tanstack query repository, the showcase scan

5,517

commits indexed

same scan

1,141

contributors mapped

same scan

0

bytes sent off the machine

loopback only, no telemetry