skills.rest
lyh2
lyh2COMMUNITY
Explore all skills by this author →
GitHub

deepseek

Analyze code diffs with review, challenge, and consult modes.

Updated Apr 3, 2026
Search Tags:#edge-cases#code-review#software-engineering#diff-analysis#brutal-honesty#ai-opinion
One-click install
npx skills add https://github.com/lyh2/gstack-deepseek --skill deepseek-lyh2
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: deepseek
Source: https://github.com/lyh2/gstack-deepseek/tree/main/deepseek
Command: npx skills add https://github.com/lyh2/gstack-deepseek --skill deepseek-lyh2

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

DeepSeek provides an independent, brutally honest second opinion on software changes, helping teams surface risks and improve review quality.

Core Features & Use Cases

  • Review mode: generate an authoritative, production-grade assessment of diffs with explicit severity markers.
  • Challenge mode: probe edge cases, adversarial inputs, and potential failure modes to strengthen robustness.
  • Consult mode: answer questions with persistent context and history, for ongoing guidance across sessions.
  • Use cases include validating PRs, stress-testing changes, and getting quick feedback on design decisions.

Quick Start

Run DeepSeek in review mode to analyze the current diff.

Frequently Asked Questions about deepseek

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get a brutally honest code review on my current git diff?▼

To get a brutally honest code review on your git diff, run the tool in review mode to analyze changes and generate a production-grade assessment with explicit severity markers and evidence. It provides an independent, verbatim response regarding code risks.

Can I stress-test edge cases and failure modes in my code changes?▼

Yes, you can stress-test edge cases and failure modes by using challenge mode to probe adversarial inputs and potential failure paths in your branch diff. This helps strengthen the robustness of your software engineering changes before merging.

What API configuration is required to run AI-powered diff analysis?▼

Running AI-powered diff analysis requires setting the DEEPSEEK_API_KEY environment variable. You can also optionally override DEEPSEEK_BASE_URL, DEEPSEEK_MODEL, and DEEPSEEK_CHAT_MODEL configurations to customize the output and routing.

Does the consult mode maintain persistent history for follow-up design questions?▼

Yes, consult mode maintains persistent context and history for ongoing guidance across sessions. It allows you to ask questions and receive continuous feedback on software engineering design decisions without losing prior conversation state.

What is the best way to validate a pull request with an independent AI opinion?▼

The best way to validate a pull request with an independent AI opinion is using review mode to analyze the current branch diff. It outputs a brutally honest assessment with severities, evidence, and token usage metrics for complete transparency.

Related Skills

Explore More Software Engineering Skills→
lyh2lyh2

gstack-deepseek-review

Run independent PR code diff reviews using the DeepSeek API.

Community
Advanced
mrp4stenmrp4sten
3

pytest

Provide structured pytest fixtures, mocking, and parametrization patterns for Python tests.

Community
Intermediate
BrennonTWilliamsBrennonTWilliams
7

verify-issue-loop

Generate finite-state machine verification loops for software issues.

Community
Advanced
ShmurdocShmurdoc

codex

Wrap the OpenAI Codex CLI for code review, adversarial testing, and consultation.

Community
Advanced