# How Many Offensive-Security Skills Are Included in Claude-Red? > Discover the 78 offensive-security skills automatically cataloged in Claude-Red's claude-skills.json manifest. Explore the capabilities of this security tool today. - Repository: [SnailSploit | Kai Aizen/Claude-Red](https://github.com/SnailSploit/Claude-Red) - Tags: getting-started - Published: 2026-09-14 --- **Claude-Red includes exactly 78 offensive-security skills**, automatically cataloged in the [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json) manifest file located at the repository root. The SnailSploit/Claude-Red repository is an open-source offensive security framework designed to extend Claude's capabilities with specialized penetration testing and red-team functionalities. Understanding the scope of **offensive-security skills included in Claude-Red** is essential for security researchers assessing the framework's coverage. The project organizes these capabilities into a machine-readable manifest that enables seamless integration with Claude's system prompt. ## Complete Inventory: 78 Offensive-Security Skills in Claude-Red According to the source code analysis of [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json) at line 602, the repository maintains a `skill_count` of **78** distinct offensive-security capabilities. This total represents the complete aggregation of skill definitions distributed across **31 categories**, ensuring comprehensive coverage of various attack vectors and methodologies. Each skill adheres to the strict `offensive-*` naming convention, ensuring immediate identification of capabilities ranging from SQL injection to advanced persistence techniques. The manifest file serves as the single source of truth for the framework's inventory, dynamically generated to prevent drift between the actual skill files and the recorded count. ## Automated Manifest Generation The framework utilizes [`tools/build_manifest.py`](https://github.com/SnailSploit/Claude-Red/blob/main/tools/build_manifest.py) to maintain accurate counts automatically. This Python script traverses the `Skills/` directory tree, parsing every [`SKILL.md`](https://github.com/SnailSploit/Claude-Red/blob/main/SKILL.md) file to compile the authoritative [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json) manifest. When executed, the build script produces output confirming the total: ```bash # Run the manifest builder to regenerate claude-skills.json python tools/build_manifest.py ``` The script outputs a summary line: ``` Wrote claude-skills.json with 78 skills across 31 categories. ``` This automation ensures that the count of **78 offensive-security skills** remains accurate as contributors add new capabilities to the repository. ## Accessing and Loading Skills Individual skill definitions reside in `Skills/**/SKILL.md` files throughout the repository. You can load these offensive capabilities into Claude using several methods depending on your operational requirements. ### Loading a Single Offensive Skill For targeted assessments, pipe a specific skill's markdown file directly into Claude's system prompt: ```bash # Display a skill's content and pipe it to Claude's system prompt cat Skills/web/offensive-sqli/SKILL.md | claude --system-file - ``` This method loads only the **offensive-sqli** skill, conserving context window space while focusing on specific vulnerability classes. ### Loading All 78 Offensive-Security Skills To initialize Claude with the complete offensive security arsenal, concatenate every [`SKILL.md`](https://github.com/SnailSploit/Claude-Red/blob/main/SKILL.md) file simultaneously: ```bash # Concatenate every SKILL.md file and feed it to Claude find Skills -name SKILL.md -exec cat {} + | claude --system-file - ``` This command aggregates all **78 offensive-security skills included in Claude-Red**, preparing the model for comprehensive red-team operations across all 31 categories. ## Summary - Claude-Red contains exactly **78 offensive-security skills** organized across 31 distinct categories. - The canonical count is stored in [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json), auto-generated by [`tools/build_manifest.py`](https://github.com/SnailSploit/Claude-Red/blob/main/tools/build_manifest.py). - Each skill follows the `offensive-*` naming convention and resides in a dedicated [`SKILL.md`](https://github.com/SnailSploit/Claude-Red/blob/main/SKILL.md) file. - Users can load individual skills or the complete inventory of 78 capabilities using standard Unix pipe commands with the Claude CLI. ## Frequently Asked Questions ### How many offensive-security skills are included in Claude-Red? Claude-Red ships with **78 offensive-security skills**. This count is recorded in the `skill_count` field within [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json), which aggregates all skill definitions from the `Skills/` directory structure. ### What is the claude-skills.json manifest? [`claude-skills.json`](https://github.com/SnailSploit/Claude-Red/blob/main/claude-skills.json) is an auto-generated machine-readable manifest located at the repository root. It catalogs every offensive-security skill in the framework, tracking the total count and metadata required to load capabilities into Claude's system prompt. ### How does Claude-Red count offensive-security skills automatically? The repository includes [`tools/build_manifest.py`](https://github.com/SnailSploit/Claude-Red/blob/main/tools/build_manifest.py), a Python script that recursively scans the `Skills/` directory, parses each [`SKILL.md`](https://github.com/SnailSploit/Claude-Red/blob/main/SKILL.md) file, and regenerates the manifest. This ensures the count of **78 skills** stays synchronized with the actual repository contents without manual updates. ### Can I load specific categories of the 78 offensive-security skills? Yes. Instead of loading all 78 skills, target specific directories within `Skills/` (such as `Skills/web/` or `Skills/network/`) using the `find` command with path constraints. This allows selective loading of offensive capabilities by category while maintaining access to the full framework inventory.