Codebase Scanning

Optional cloud analysis that bootstraps structural context from a GitHub repository — features, tech stack, and file relationships.

Overview

Scanning is an optional step that bootstraps structural context from a GitHub repository. It maps features, tech stack, architecture layers, and code relationships — giving your AI a foundation to build on before you have written a single memory entry.

Not required to start

Remb does not require a scan. Context accumulates from memories, feature notes, and conversation logs from your very first session. Scanning is additive — it supplements the context you build manually, it does not replace it.

How It Works

When you trigger a scan, Remb clones or reads your GitHub repository and runs a multi-phase analysis pipeline. Results are written back to your project as structured feature context — queryable by your AI on demand.

Scans are processed asynchronously. You can monitor progress in real time via the dashboard or the CLI.

Triggering a Scan

There are three ways to trigger a scan:

Method	Command / Action	Notes
`CLI — manual`	`remb scan`	`Runs immediately against the linked GitHub repo.`
`CLI — on push`	`remb push`	`Triggers a scan after verifying your git state. Suitable for post-deploy hooks.`
`Dashboard`	`Projects → Scan`	`One-click scan from the project overview page.`

CLI examples

# Trigger a scan
remb scan

# Trigger after a git push (with pre-flight checks)
remb push

# Stream live scan progress
remb status

Scan Phases

Every scan runs through five sequential phases:

Phase	Name	What It Does
`1`	`Scout`	`Reads the repository structure, identifies languages, frameworks, and entry points.`
`2`	`Analyze`	`Parses source files to extract features, modules, and code symbols.`
`3`	`Architect`	`Maps relationships between components — imports, call graphs, dependency layers.`
`4`	`Review`	`Validates extracted data for consistency and fills in gaps with inference.`
`5`	`Finalize`	`Writes structured context entries and feature records to your project.`

What Gets Extracted

After a completed scan, your project will contain:

A list of detected features with descriptions and key files
Tech stack summary — languages, frameworks, databases, and tooling
Architecture layer map — frontend, API, services, and data layers
Code symbol index — functions, classes, and components with relationships
Dependency graph — inter-feature and inter-module links

All of this is queryable by your AI via MCP tools — no additional setup required.

Monitoring Progress

You can monitor a running scan in real time:

# Stream live progress in the terminal
remb status

The dashboard also shows scan status, phase progress, and a log feed under Projects → Scans.