Qiushi Skill
求是 Skill
A set of methodology skills: investigate the facts first, find the main contradiction, validate in practice and finish the job
- 3,861 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The set consists of instructions, methodology skills and prompts
- A session-start hook injects the kernel text without performing dangerous operations
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/hughyau-qiushi-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The set consists of instructions, methodology skills and prompts
- A session-start hook injects the kernel text without performing dangerous operations
This is third-party code. Review the repository files before installing.
What it does
The set gives an agent a method for working through problems, assembled from classic materialist dialectics and the philosophy of practice, drawing on Mao Zedong's essays. At its core is one governing principle, seek truth from facts, and nine methodological tools: contradiction analysis, practice and cognition, field investigation, the mass line, criticism and self-criticism, protracted strategy, concentration of forces, spark-starts-a-fire, and overall planning, plus a workflow layer that chains them. At session start a small arming-thought kernel loads and sets four hard rules (conclusions follow evidence, separate fact, inference and unknown, count only verified work as done, diagnose before giving up) and pulls in a specific skill only when it clearly helps. Each skill is written in five parts: when to use, procedure, output template, discipline and handoff, while source quotations sit in original-texts.md and are not auto-loaded. It ships two subagents (an investigator and a self-critic), manual commands and a session-start hook.
Who it is for. For anyone who wants an agent that investigates and verifies before concluding: developers, analysts, researchers and managers.
Good fit when
- You need an agent that gathers facts first and only then draws conclusions
- The task is complex and you need to find the main point and set priorities
- You have a hypothesis or solution and need to validate it in practice and weed out loose ends
- The task is long and needs a strategy, not a one-off answer
Not a fit when
- A simple direct task where the methodology would only slow things down
- The host already provides plan, review and todo flows and there is no need to duplicate them
Example request
Work through this complex task by the methodology: gather the facts first, then find the main contradiction and propose a planLimitations
The skill kernel and the source quotations are in Chinese, though there is an English README. It is a working methodology, not fact-checking and not a guarantee of a correct answer. Installing via npx needs Node.js 18.17+; without it the folders from skills/ are copied by hand. No keys or paid accounts are required.
How to disable. Remove the set with npx qiushi-skill uninstall --target <host>, the plugin in Claude Code via /plugin, or just delete the copied skill folders from the host's directory.
FAQ
What language is it in?
The skill kernel and the source quotations are in Chinese; there is an English README.
Is it a political project?
The author frames the set as a methodology for analyzing and solving problems, not as politics or propaganda.
Discussion
No questions or reviews yet.
Related
SkillsA skill that cross-examines a plan before work starts: numbered questions in rounds with a recommended answer and no changes until you confirm
Four skills for a person with several AI assistants: a task register, evidence-based acceptance, beginner instructions
A collection of skills for decisions and projects: strategy, market research, thinking, data analysis, docs and slides.
Twenty-seven Chinese-language skills for development, thinking, learning, writing, visuals, branding and working with agents
Ten ShareAI Lab skills: source-level architecture research, review of past agent sessions, meeting coaching and article adaptation
Nuwa
女娲.skill (Nuwa)
A skill that researches how any person thinks and turns it into a ready-to-use advisor skill
README in short
The README presents the set as a collection of methodology skills for agents, assembled from classic dialectics and the philosophy of practice with direct source quotations. It describes the governing principle of seeking truth from facts, a three-tier structure of nine tools, and an entry kernel that sets the rules at session start and routes to the right skill. Installation is via npx qiushi-skill or the Claude Code marketplace, and without Node.js the folders are copied by hand. It also lists subagents, manual commands and supporting files. MIT licensed.
README badge
Original SKILL.md
--- name: arming-thought description: | 求是方法论的入口与总原则。在每次新的顶层会话开始时加载一次，用「实事求是」约束全部判断，并只在明确适用时把任务路由到对应的方法论 skill；被派遣执行单一具体任务的子 agent 直接跳过。 English: Entry point and governing principle of the Qiushi methodology. Load once at the start of a top-level session to enforce "seek truth from facts" and route to a downstream skill only when it clearly helps. Delegated sub-agents on narrow tasks should skip it. --- # 武装思想 > "'实事'就是客观存在着的一切事物，'是'就是客观事物的内部联系，即规律性，'求'就是我们去研究。" —— 《改造我们的学习》 **子 agent 执行单一具体任务时：跳过本 skill。** ## 总原则：实事求是（四条硬规则） | 规则 | 做法 | 违反信号 | |---|---|---| | 结论跟着证据走 | 每个判断标出依据：读了什么、跑了什么、看到了什么 | 有判断，无依据 | | 分清三种话 | 区分「已核实的事实」「推断」「未知」；未知就写"不知道，需确认" | 用猜测填空 | | 验证过才算完成 | 声称完成前先跑一遍、看输出、比对预期，并写出验证动作 | 写完即宣布完成 | | 遇阻先诊断 | 失败时先说清原因，再补调查或换路径；不在第一次受阻就停 | "超出能力范围"式放弃 | 事实与预想冲突时，改的是判断，不是事实。