GitHub Audit
Audit a GitHub repository for code quality, dependency risk, and best practices.
A GitHub audit clones a repository, reads its structure, and produces a code-quality report across the same scoring dimensions as a URL audit — with the content and GEO categories adapted for source-code analysis.
When to use
- Evaluating a new codebase before joining a project
- Auditing a client's repository as part of a discovery phase
- Checking dependency risk across a monorepo
- Establishing a code-quality baseline before a refactor
How to run
Open Run audit
From your project page, click Run audit.
Select GitHub
Choose GitHub from the audit type options.
Enter repository URL
Paste the full GitHub URL (e.g.
https://github.com/acme/website). Public repositories work without any credentials.Submit
Click Start audit. The worker clones the repository, reads the file tree, and runs analysis.
What it measures
The GitHub audit uses three specialist tools:
- Clone tool — shallow clone of the default branch; reads file tree,
package.json/requirements.txt/Gemfile, CI config files. - Read tool — samples key source files to assess code structure and documentation quality.
- Dependency scan — checks declared dependencies against known vulnerability databases; flags outdated major versions.
What it returns
- Best Practices score — code structure, documentation, CI/CD presence, linting configuration
- Dependency risk report — list of packages with known vulnerabilities or major-version lag
- Recommendations — specific file-level suggestions (e.g. "Add
.nvmrcfor Node version pinning", "Upgrade lodash from 4.17.15 to 4.17.21")
GitHub audit requires Anthropic BYOK
The GitHub audit agent uses Claude via the Anthropic API directly. It requires either the system Anthropic key (included on all plans) or your own key configured in Settings → API Keys. OpenAI BYOK does not support GitHub audits.
Private repositories
Private repo access is not yet supported in the hosted SitePulse product. The worker can only clone repos that respond to a public HTTPS clone without credentials.
Processing time
| Repository size | Typical time |
|---|---|
| Small (< 100 files) | 45–90 seconds |
| Medium (100–500 files) | 90–180 seconds |
| Large (> 500 files) | 2–5 minutes |