Testing Handbook Skills - Trail of Bits Skills

Meta-skill that generates Claude Code skills from the Trail of Bits Application Security Testing Handbook, providing expert guidance on security testing tools and techniques.

Overview

This plugin contains a skill generator that analyzes the Testing Handbook structure and produces individual skills for security testing tools, fuzzing techniques, and testing methodologies. Each generated skill provides:

Tool-specific guidance - Setup, configuration, and best practices
Language-specific patterns - Tailored for C/C++, Rust, Python, Go, etc.
Practical examples - Real-world usage patterns from security audits
Integration patterns - CI/CD integration and workflow automation

The plugin includes both the meta-skill generator AND 16 pre-generated skills covering fuzzers, static analysis tools, and testing techniques.

Generated Skills

Fuzzers (6)

libFuzzer

Coverage-guided fuzzing for C/C++ with LLVM integration

AFL++

Multi-core fuzzing with advanced instrumentation

libAFL

Modular fuzzing framework in Rust

cargo-fuzz

Rust-native fuzzing with libFuzzer backend

Atheris

Coverage-guided fuzzing for Python code

Ruzzy

Fuzzing for Ruby applications

Techniques (6)

Harness Writing

Patterns for effective fuzz target construction

Address Sanitizer

Memory error detection during testing

Coverage Analysis

Measuring and improving test coverage

Fuzzing Dictionary

Custom dictionaries for protocol fuzzing

Fuzzing Obstacles

Overcoming checksums, magic bytes, and complex parsers

OSS-Fuzz

Continuous fuzzing for open-source projects

Static Analysis (2)

Semgrep

Fast pattern-based static analysis

CodeQL

Query-based semantic code analysis

Domain-Specific (2)

Wycheproof

Cryptographic library testing suite

Constant-Time Testing

Timing side-channel verification

Skill Cross-Reference

Generated skills reference each other based on natural relationships: Legend:

Solid arrows: Primary dependencies (techniques/tools used together)
Dashed arrows: Alternatives (similar tools/fuzzers)

Generator Usage

Full Handbook Generation

Generate skills from all applicable handbook sections:

# Claude will:
# 1. Locate the handbook (or clone from GitHub)
# 2. Analyze handbook structure
# 3. Present generation plan
# 4. Generate skills on approval
# 5. Validate output
# 6. Update documentation

Workflow:

Locate handbook (check common locations, ask user, or clone)
Read discovery.md methodology
Scan handbook at {handbook_path}/content/docs/
Build candidate list with types (Tool/Fuzzer/Technique/Domain)
Present plan to user
On approval, generate each skill using appropriate template
Validate generated skills with validate-skills.py
Update main README with skills table
Update cross-reference graph

Single Section Generation

Generate a skill from a specific handbook section:

"Create a skill for the libFuzzer section of the testing handbook"

The generator will:

Read /testing-handbook/content/docs/fuzzing/c-cpp/10-libfuzzer/
Identify type: Fuzzer Skill
Apply fuzzer template
Extract content and examples
Write to skills/libfuzzer/SKILL.md
Validate and report

Regeneration

To update an existing skill after handbook changes:

"Regenerate the libfuzzer skill from the latest handbook content"

Skill Types and Templates

The generator uses four skill templates:

Type	Template	Example Sources	Key Sections
Tool	`tool-skill.md`	Semgrep, CodeQL	Installation, Configuration, Patterns, CI Integration
Fuzzer	`fuzzer-skill.md`	libFuzzer, AFL++	Setup, Harness, Corpus, Options, Sanitizers
Technique	`technique-skill.md`	Harness writing, Coverage	Methodology, Patterns, Best Practices, Examples
Domain	`domain-skill.md`	Wycheproof, Constant-time	Domain Context, Tools, Test Suites, Validation

Quality Validation

Generated skills are validated with scripts/validate-skills.py:

# Validate all skills
uv run scripts/validate-skills.py

# Validate specific skill
uv run scripts/validate-skills.py --skill libfuzzer

# JSON output for CI
uv run scripts/validate-skills.py --json

Validation checks:

Valid YAML frontmatter with name and description
Required sections present (When to Use, When NOT to Use)
Line count under 500 (split into references/ if needed)
Cross-references to existing skills only
No broken internal links

Generator Architecture

Two-Pass Generation

Solves forward reference problem (skills referencing skills that don’t exist yet): Pass 1: Content Generation (Parallel)

Generate all skills simultaneously
Skip “Related Skills” section
Write content to skills/{skill-name}/SKILL.md

Pass 2: Cross-Reference Population (Sequential)

Read all generated skill names
Determine related skills based on handbook structure
Update each SKILL.md with Related Skills section
Validate all cross-references exist

Agent Prompt Template

See skills/testing-handbook-generator/agent-prompt.md for full template with:

Variable substitution (skill name, type, handbook path, pass number)
Pre-write validation checklist
Hugo shortcode conversion to Mintlify components
Line count splitting rules
Output report format

Handbook Location

The generator automatically:

Checks common locations: ./testing-handbook, ../testing-handbook, ~/testing-handbook
Asks user for path if not found
Clones from GitHub as last resort: https://github.com/trailofbits/testing-handbook

No hardcoded paths - adapts to your environment.

Example Generated Content

Each generated skill includes:

Frontmatter - Name, description for skill activation
When to Use - Concrete triggers and use cases
When NOT to Use - Antipatterns and wrong scenarios
Quick Start - Installation and basic usage
Configuration - Tool-specific setup and options
Examples - Real-world patterns from security testing
CI Integration - GitHub Actions/GitLab CI examples
Troubleshooting - Common issues and solutions
Related Skills - Cross-references to complementary skills
Resources - External documentation (WebFetch summaries, no video content)

Self-Improvement

After each generation run, the generator captures improvements: Improvement categories:

Template enhancements (missing sections, better structure)
Discovery logic updates (pattern detection, false positives)
Content extraction fixes (shortcode handling, formatting)

Update process:

Note issues during generation
Identify patterns that caused problems
Update relevant files:
- SKILL.md - Workflow and decision tree
- templates/*.md - Template improvements
- discovery.md - Detection logic
- testing.md - Validation checks

Integration with Other Plugins

Property-Based Testing - Combine fuzzing with property tests
Constant-Time Analysis - Use with constant-time testing skill
Zeroize Audit - Memory safety testing patterns
Spec-to-Code Compliance - Verify test coverage against spec

Resources