Back to directory

Qiushi Skill

An Agent Skills collection that gives an AI agent nine methodological tools drawn from classical materialist dialectics and practice philosophy, for investigating a problem, focusing on its main contradiction and pushing work through to completion.

Qiushi Skill

Qiushi-Skill is a collection of Agent Skills that packages one guiding principle and nine methodological tools, drawn from classical materialist dialectics and practice philosophy, into a systematic reasoning process for an AI agent. Its stated goal is to address agents that reason but do not 'think through' a problem: jumping to answers without investigation, failing to self-check finished work, or giving up when a task looks too hard.

The guiding principle, seeking truth from facts, constrains every judgment; the nine tools are organized in three layers: a philosophical base of contradiction analysis and practice-based epistemology, working methods of investigation and mass-line feedback and self-criticism, and strategic and tactical tools including protracted strategy, concentrating superior force, starting small and building outward, and balancing multiple priorities. Each method cites the original classical text it is drawn from, and a companion set of cross-skill workflows defines how methods are chained together with data passed between them.

The author states this is a methodology collection, not a political or propaganda project, and not an attempt to simulate a person's personality.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.