howardpen9

Deepseek Skill — Security skill for Claude Code

Security community

DeepSeek V4 as a Claude Code skill — five modes including 'audit' that challenges Claude's reasoning in the current session.

How to install Deepseek Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open howardpen9/deepseek-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Deepseek Skill does

DeepSeek V4 as a Claude Code skill — five modes including 'audit' that challenges Claude's reasoning in the current session.

Alternatives in Security

Octoboss Reorganize Tentacles — You are the Octoboss — a cross-tentacle orchestrator 1.4k ★
Maestro Odyssey — Long-running iterative cycle — one entry, seven modes (debug improve planex review security defensive ui) 530 ★
Release Audit — Use right before cutting a release — it spawns ONE fresh sibling session via CCC that audits only the JUDGMENT 130 ★

README

deepseek-skill

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) [![GitHub stars](https://img.shields.io/github/stars/howardpen9/deepseek-skill?style=social)](https://github.com/howardpen9/deepseek-skill) [![Last Commit](https://img.shields.io/github/last-commit/howardpen9/deepseek-skill)](https://github.com/howardpen9/deepseek-skill/commits/main) [![Issues](https://img.shields.io/github/issues/howardpen9/deepseek-skill)](https://github.com/howardpen9/deepseek-skill/issues)

Turn DeepSeek V4 into a third-party reviewer for your Claude Code session — including an `audit` mode that challenges your current AI's reasoning to surface bias, hallucinations, and unstated assumptions.

A single-file Claude Code skill that talks to the DeepSeek API directly via `curl + jq`. No MCP server. No npm package. No CLI wrapper. Drop in, set one env var, and call `/deepseek ` from any session.

繁體中文版：[README.zh-TW.md](./README.zh-TW.md)

**Contents:** [Why](#why-this-exists) · [Modes](#what-you-get--five-sub-modes) · [Install](#install) · [Usage](#usage) · [Comparison](#how-this-compares-to-other-deepseek-tools) · [FAQ](#faq) · [Real audit example](#real-audit-example) · [Roadmap](#roadmap) · [Contributing](#contributing)

Why this exists

DeepSeek already has good MCP servers ([DMontgomery40/deepseek-mcp-server](https://github.com/DMontgomery40/deepseek-mcp-server), [arizen-dev/deepseek-mcp](https://github.com/arizen-dev/deepseek-mcp)). This isn't one of them.

What this **is**: a prompt-engineering layer on top of the DeepSeek API, packaged as a Claude Code skill, with one feature the MCP servers don't have — **`audit` mode**, which feeds Claude's own current reasoning back to DeepSeek and asks DeepSeek to find what Claude got wrong.

It's a meta-cognitive layer for AI coding sessions.