Skip to main content

​
audit-website Skill

The audit-website skill enables AI agents to perform comprehensive website audits using the squirrelscan CLI. It checks 230+ rules across 21 categories including SEO, performance, security, accessibility, and content quality.

​
What This Skill Does

This skill gives agents the ability to:
  • Audit websites against 230+ rules in 21 categories
  • Generate LLM-optimized reports with health scores and recommendations
  • Detect broken links (internal and external)
  • Analyze meta tags and structured data
  • Identify technical issues like redirect chains and page speed problems
  • Check security for leaked secrets, HTTPS usage, and security headers
  • Validate accessibility including alt text and color contrast
  • Compare audits to detect regressions over time

​
Audit Categories

The skill covers 21 comprehensive categories:
  • Meta tags, titles, descriptions
  • Canonical URLs
  • Open Graph tags
  • Keyword optimization
  • Content structure
  • Broken links
  • Redirect chains
  • Page speed
  • Mobile-friendliness
  • Crawlability
  • Page load time
  • Resource usage
  • Caching strategies
  • Image optimization
  • Leaked secrets
  • HTTPS usage
  • Security headers
  • Mixed content
  • Heading structure (H1-H6)
  • Image alt text
  • Content analysis
  • E-E-A-T signals
  • Accessibility
  • Usability
  • Mobile optimization
  • Schema markup
  • Legal compliance
  • Social media tags
  • Local SEO
  • Video optimization

​
When to Use This Skill

Use the audit-website skill when you need to:
  • Analyze overall website health
  • Debug technical SEO issues
  • Find and fix broken links
  • Validate meta tags and structured data
  • Generate comprehensive audit reports
  • Compare site health before and after changes
  • Improve performance, accessibility, or security
  • Prepare for a product launch
  • Monitor website quality over time
Re-audit frequently to ensure your website remains healthy. The skill is designed for iterative improvement.

​
Installation

​
Prerequisites

This skill requires the squirrel CLI to be installed and accessible in your PATH.
The skill will not work without squirrel installed. Follow the installation steps below.

​
macOS and Linux Installation

1

Install squirrel

Run the installation script:
curl -fsSL https://squirrelscan.com/install | bash
This will:
  • Download the latest binary
  • Install to ~/.local/share/squirrel/releases/{version}/
  • Create a symlink at ~/.local/bin/squirrel
  • Initialize settings at ~/.squirrel/settings.json
2

Add to PATH

If ~/.local/bin is not in your PATH, add it:
# Add to ~/.bashrc or ~/.zshrc
export PATH="$HOME/.local/bin:$PATH"

# Reload your shell
source ~/.bashrc  # or source ~/.zshrc
3

Verify Installation

Check that squirrel is working:
squirrel --version
You should see version information printed.

​
Windows Installation

1

Run PowerShell Installer

Open PowerShell and run:
irm https://squirrelscan.com/install.ps1 | iex
This will:
  • Download the latest Windows binary
  • Install to %LOCALAPPDATA%\squirrel\
  • Add squirrel to your PATH
2

Restart Terminal

Restart your terminal for PATH changes to take effect
3

Verify Installation

Check that squirrel is working:
squirrel --version

​
Configuration

​
Project Setup

Before auditing, initialize a squirrel project: 
# Initialize with a project name
squirrel init --project-name my-website

# Or use short form
squirrel init -n my-website

# Force overwrite existing config
squirrel init -n my-website --force
This creates a squirrel.toml configuration file.
The project name identifies the audit database. All audits for a project share the same database, enabling comparison over time.

​
Configuration File

The squirrel.toml file controls audit behavior:
squirrel.toml
[project]
name = "my-website"

[crawl]
max_pages = 200
coverage = "surface"  # quick | surface | full

[audit]
min_score = 85

​
Project Database

Audits are stored in: ~/.squirrel/projects/<project-name> This enables:
  • Historical comparison
  • Regression detection
  • Report regeneration in different formats

​
Usage

​
Basic Audit Workflow

The standard audit process: 
# Quick surface scan (100 pages)
squirrel audit https://example.com --format llm
Always use --format llm for AI agent workflows. This format is optimized for token efficiency and provides exhaustive, structured output.

​
Coverage Modes

Choose the right coverage for your needs:
ModePagesBehaviorUse Case
quick25Seed + sitemaps onlyCI checks, fast monitoring
surface100One sample per URL patternGeneral audits (default)
full500Crawl everythingDeep analysis, pre-launch
Surface mode intelligence: It detects URL patterns like /blog/{slug} and only crawls one sample per pattern. This makes it efficient for sites with many similar pages.

​
Output Formats

The skill supports multiple formats: 
# LLM-optimized (best for agents)
squirrel audit https://example.com --format llm

# Human-readable console
squirrel audit https://example.com --format console

# JSON for processing
squirrel audit https://example.com --format json

# HTML report
squirrel audit https://example.com --format html --output report.html
Agents should always use --format llm. It’s 40% more compact than XML and optimized for AI consumption.

​
Workflow: Audit → Fix → Re-Audit

The skill follows a systematic improvement workflow:
1

Initial Audit

Run a surface scan to identify issues:
squirrel audit https://example.com --format llm
This generates:
  • Overall health score (0-100)
  • Category breakdowns
  • Specific issues with affected URLs
  • Actionable recommendations
2

Fix Issues

Address all critical errors and warnings:
  • Code fixes: Meta tags, structured data, templates
  • Content fixes: Alt text, headings, descriptions
  • Technical fixes: Broken links, redirects, performance
Don’t stop after code fixes. Content changes (*.md, *.mdx) are equally important.
3

Re-Audit

Verify improvements with a fresh audit:
squirrel audit https://example.com --refresh --format llm
The --refresh flag ignores cache to ensure accurate results.
4

Iterate Until Complete

Continue fixing and re-auditing until:
  • Score reaches target (typically 85-95+)
  • Only issues requiring human judgment remain
A site is only considered complete and fixed when scores are above 95 (Grade A) with --coverage full.

​
Score Targets

Set improvement goals based on starting score:
Starting ScoreTarget ScoreExpected Work
< 50 (Grade F)75+ (Grade C)Major fixes required
50-70 (Grade D)85+ (Grade B)Moderate fixes needed
70-85 (Grade C)90+ (Grade A)Polish and refinement
> 85 (Grade B+)95+ (Grade A+)Fine-tuning
Don’t stop until the target is reached.

​
Parallelizing Fixes with Subagents

The skill leverages subagents to fix issues in parallel, dramatically reducing completion time.

​
When to Parallelize

Parallelize when:
  • 5+ files need the same fix type
  • Fixes have no dependencies on each other
  • Files are independent (not importing each other)

​
Common Parallelizable Fixes

Issue TypeParallelizableApproach
Image alt text✅ YesSpawn subagents per file batch
Heading hierarchy✅ YesSpawn subagents per file batch
Short descriptions✅ YesSpawn subagents per file batch
HTTP→HTTPS links✅ YesBulk sed/replace
Meta tags/titles❌ NoShared components
Structured data❌ NoSingle source of truth
Broken links❌ NoRequires manual review

​
Parallel Execution Pattern

Multiple Task tool calls in ONE message = parallel execution. Sequential calls = slower.
Example: Fixing alt text in 12 files Spawn 3 subagents in a single message: 
Fix missing image alt text in these files:
- content/blog/post-1.md
- content/blog/post-2.md
- content/blog/post-3.md
- content/blog/post-4.md

Find images without alt text (![](path) or <img without alt=).
Add descriptive alt text based on image filename and context.
Do not ask for confirmation.

​
Batch Sizing Guidelines

  • Optimal: 3-5 files per subagent
  • Maximum: 10 files per subagent
  • Total agents: Spawn 2-4 subagents for parallel work

​
Subagent Prompt Structure

Effective subagent prompts are:
  1. Focused - Specific file list
  2. Clear - Exact pattern to find
  3. Actionable - Precise fix instructions
  4. Autonomous - “Do not ask for confirmation”
Fix [issue type] in the following files:
- path/to/file1.md
- path/to/file2.md

Pattern: [what to find]
Fix: [what to change]

Do not ask for confirmation. Make all changes and report what was fixed.

​
Advanced Options

​
Custom Page Limits

# Audit more pages than default
squirrel audit https://example.com --max-pages 200

# Maximum allowed
squirrel audit https://example.com --max-pages 5000

​
Force Fresh Crawl

# Ignore cache, fetch all pages fresh
squirrel audit https://example.com --refresh

​
Resume Interrupted Crawl

# Continue from where it stopped
squirrel audit https://example.com --resume

​
Debugging

# Verbose output
squirrel audit https://example.com --verbose

# Debug logging
squirrel audit https://example.com --debug

# Performance tracing
squirrel audit https://example.com --trace

​
Regression Detection

Compare audits to detect regressions: 
# Compare against a baseline audit ID
squirrel report --diff <audit-id> --format llm

# Compare against a baseline domain
squirrel report --regression-since example.com --format llm
Diff mode works with: console, text, json, llm, markdown
Diff mode does not support html or xml formats.

​
Complete Workflow Example

A real-world audit and fix workflow:
1

Setup

# Initialize project
cd ~/projects/my-saas
squirrel init -n my-saas
2

Initial Audit

# Surface audit of production site
squirrel audit https://my-saas.com --format llm
Results:
  • Score: 68/100 (Grade D)
  • 43 errors, 89 warnings
  • Categories: SEO issues, missing alt text, heading problems
3

Plan Fixes

Identify parallelizable work:
  • 12 files missing alt text → 3 subagents
  • 18 files with heading hierarchy issues → 3 subagents
  • 7 files with short descriptions → 1 subagent
  • Meta tag fixes → Main agent (shared components)
4

Execute Fixes

Spawn subagents in parallel for content fixes. Main agent handles shared component updates.
5

Re-Audit

squirrel audit https://my-saas.com --refresh --format llm
Results:
  • Score: 87/100 (Grade B)
  • 4 errors, 12 warnings
6

Final Round

Fix remaining issues.
squirrel audit https://my-saas.com --refresh --format llm
Results:
  • Score: 96/100 (Grade A+)
  • 0 errors, 2 warnings (require human judgment)

​
Completion Criteria

A site audit is complete when:
  • ✅ All errors fixed
  • ✅ All warnings fixed or documented as requiring human review
  • ✅ Re-audit confirms improvements
  • ✅ Before/after comparison shown
  • ✅ Score above 95 with full coverage
Don’t stop early. Fix ALL issues, not just the obvious ones. Continue until score targets are met.

​
Troubleshooting

​
Command Not Found

Problem: squirrel: command not found Solution: 
# 1. Install squirrel
curl -fsSL https://squirrelscan.com/install | bash

# 2. Add to PATH
export PATH="$HOME/.local/bin:$PATH"

# 3. Verify
squirrel --version

​
Permission Denied

Problem: Permission error when running squirrel Solution: 
chmod +x ~/.local/bin/squirrel

​
Invalid URL

Problem: Audit fails with URL error Solution: Include protocol in URL 
# ✗ Wrong
squirrel audit example.com

# ✓ Correct
squirrel audit https://example.com

​
Slow Performance

Problem: Audit takes too long Solution: Use quick coverage or reduce max pages 
# Fast health check
squirrel audit https://example.com --coverage quick --format llm

# Or limit pages
squirrel audit https://example.com --max-pages 50 --format llm

​
Additional Resources

squirrelscan Docs

Full documentation including rule references

Rule Reference

Detailed explanations of all 230+ rules

CLI Help

Run squirrel audit --help for command reference

Website

squirrelscan homepage and updates

​
Rule Documentation

Look up specific rules: 
https://docs.squirrelscan.com/rules/{category}/{rule_id}
Example: 
https://docs.squirrelscan.com/rules/links/external-links

​
Report Commands

# List recent audits
squirrel report --list

# Generate report from audit ID
squirrel report <audit-id> --format llm

# Filter by severity
squirrel report <audit-id> --severity error

# Filter by categories
squirrel report <audit-id> --category seo,performance

​
Config Commands

# Show current config
squirrel config show

# Set config value
squirrel config set project.name my-site

# Show config file path
squirrel config path

# Validate config
squirrel config validate

​
Self-Management Commands

# Update squirrel
squirrel self update

# Health check
squirrel self doctor

# Version info
squirrel self version

# Shell completions
squirrel self completion

​
Best Practices

  1. Always use LLM format for agent workflows
  2. Start with surface audits before full scans
  3. Re-audit after fixes to verify improvements
  4. Parallelize content fixes using subagents
  5. Don’t stop early - reach score targets
  6. Audit frequently to catch issues early
  7. Test before deployment by auditing staging/preview environments

Build docs developers (and LLMs) love

Get started for freeTalk to us