Audit Codebase

Help me understand the codebase for: $ARGUMENTS

Instructions

Conduct a PM-focused codebase exploration. The goal is not to understand every line of code, but to map what the system can do and what that means for the product.

Process

1. Auto-detect structure — Identify language, framework, repo layout, entry points
2. Explore architecture — Map services, components, data flows, and external dependencies
3. Map capabilities — What product capabilities does this code enable?
4. Find the gaps — What's claimed in docs but missing in code? What exists but isn't documented?
5. Connect to product decisions — What does this mean for what we can build, change, or promise?

Tools to Use

• Search the codebase for key patterns, APIs, and data models
• Read configuration files, READMEs, and service definitions
• Trace user-facing flows from entry point to data store
• Compare implementation reality to documentation claims
• GitHub MCP (if available): Pull open issues, recent PRs, contributor activity for context on what's changing
• Sentry MCP (if available): Check for recurring errors or performance issues in this service

Auto-Detection

Start by detecting the repo's characteristics:

## Repo Profile
- **Language(s):** [auto-detect from file extensions and config]
- **Framework:** [detect from package.json, pom.xml, build.gradle, etc.]
- **Architecture:** [monolith / microservice / monorepo / library]
- **Entry points:** [main files, server startup, route definitions]
- **Data stores:** [databases, caches, queues detected from config/code]
- **External deps:** [APIs, services this code calls]

Output Format

# Codebase Audit: [System/Service Name]

## One-Sentence Summary
What this system does in plain language.

## Repo Profile
| Attribute | Value |
|-----------|-------|
| Language | [detected] |
| Framework | [detected] |
| Architecture | [type] |
| Size | [files/lines estimate] |

## Architecture Overview

[Mermaid diagram showing key components, data flow, and external dependencies]

## Capability Map
…

Principles

• Evidence over assumption — Cite file paths and code patterns
• Product language — Translate technical findings into PM-relevant insights
• Honest about limits — Flag what you can't determine from code alone
• Mermaid for diagrams — Use Mermaid format for architecture diagrams, not ASCII art
• Auto-detect first — Don't ask the user for repo details you can discover by reading the code