DevOpsscout
Dark Code Auditor
Find code nobody understands in your GitHub repo
What it does
Dark Code Auditor analyzes GitHub repos for code no one has written, read, or understood. Connect GitHub, pick a repo, and get a 0–100 score with per-module breakdown. Heuristic scanning covers documentation, types, error handling, test coverage, and AI-generated code. AI-augmented mode adds deeper Gemini-powered analysis.
The Problem
AI-generated and copy-pasted code accumulates silently until nobody can own it.
- No visibility into which modules contain unreviewed AI-generated code
- Technical debt accumulates below the radar of traditional linters
- New hires inherit a codebase they can't reason about
- M&A due-diligence has no way to quantify code comprehension risk
The Solution
Heuristic + AI-augmented scanning that scores every module for comprehension risk.
- 0-100 Dark Code score with per-module drill-down
- Heuristic checks: docs, types, error handling, test coverage
- AI re-evaluation of ambiguous files via Gemini
- Trend tracking to measure improvement over time
Features
- GitHub Connection: Connect via OAuth and scan both private and public repos.
- Heuristic Scanning: Automated analysis of documentation, type declarations, error handling and test coverage.
- AI-Augmented Analysis: Gemini-powered re-evaluation of ambiguous files and AI-generated code detection.
- Per-Module Breakdown: View Dark Code score per folder with drill-down to individual files.
- Trend Tracking: Compare scans over time to measure improvement.
- Privacy Mode: Upload a CLI manifest without granting GitHub access.
Use cases
- DevOps teams: Identifying technical debt and unknown code in legacy systems and CI pipelines.
- CTOs: Measuring code comprehension across the team to spot bus-factor risks.
- Security teams: Finding unknown or AI-generated code in production before attackers do.
- Onboarding: New developers find code that needs documentation before shipping changes.
- M&A due diligence: Technical assessment of Dark Code risk when acquiring tech companies.
FAQ
- What is Dark Code?
- Code in production that no one fully understands — AI-generated without review, copied from Stack Overflow, or written by a developer who has left.
- Is my code safe?
- Yes. GitHub tokens are stored encrypted in Supabase Vault. We only read file content for analysis and delete everything after 30 days (Scout) or 12 months (paid plans).
- What does it cost?
- 5 credits for heuristic scanning, 15 credits for AI-augmented analysis.
- Which languages are supported?
- TypeScript/JavaScript, Python, Java, Go, and C# in V1.
Try Dark Code Auditor free