How hughyau/qiushi-skill works
Qiushi Skill is a collection of practical methodology skills for AI agents, with a session-start kernel that routes tasks to the appropriate methods and a separate CLI for installing those assets into supported hosts. The documented reasoning workflow starts from evidence, selects relevant methods, and can combine them through workflows; a manual command directory provides alternate entry points. The graph preserves the nine distinct methods and major integration paths. CLI wiring is based on sampled source; method-level relationships reflect the README, since skill instructions were not sampled.
Main entry points: arming-thought/SKILL.md Arming Kernel · commands Slash Commands · workflows/SKILL.md Workflow Compositions · qiushi-skill.mjs CLI Entry
Components 20
Methodology Skills
- Contradiction Analysis
skills/contradiction-analysis/SKILL.md - Practice Cognition
skills/practice-cognition/SKILL.md - Investigation First
skills/investigation-first/SKILL.md - Mass Line
skills/mass-line/SKILL.md - Self-Criticism
skills/criticism-self-criticism/SKILL.md - Protracted Strategy
skills/protracted-strategy/SKILL.md - Concentrate Forces
skills/concentrate-forces/SKILL.md - Spark Prairie Fire
skills/spark-prairie-fire/SKILL.md - Overall Planning
skills/overall-planning/SKILL.md
Agent Entry Points
- Arming Kernel
skills/arming-thought/SKILL.md - Slash Commands
commands - Session Hooks
hooks - Subagents
agents - Workflow Compositions
skills/workflows/SKILL.md
Installation CLI
- CLI Entry
bin/qiushi-skill.mjs - Platform Detection
bin/lib/detect-platform.mjs - Asset Installer
bin/lib/install.mjs
Other components
- User
- AI Agent
- Agent Hosts
Connections 37
- User → AI Agent: requests work
- AI Agent → Arming Kernel: applies kernel
- Session Hooks → Arming Kernel: injects at session start
- Arming Kernel → Contradiction Analysis: routes when relevant
- Arming Kernel → Practice Cognition: routes when relevant
- Arming Kernel → Investigation First: routes when relevant
- Arming Kernel → Mass Line: routes when relevant
- Arming Kernel → Self-Criticism: routes when relevant
- Arming Kernel → Protracted Strategy: routes when relevant
- Arming Kernel → Concentrate Forces: routes when relevant
- Arming Kernel → Spark Prairie Fire: routes when relevant
- Arming Kernel → Overall Planning: routes when relevant
- AI Agent → Slash Commands: invokes manually
- Slash Commands → Contradiction Analysis: loads matching guidance
- Slash Commands → Practice Cognition: loads matching guidance
- Slash Commands → Investigation First: loads matching guidance
- Slash Commands → Mass Line: loads matching guidance
- Slash Commands → Self-Criticism: loads matching guidance
- Slash Commands → Protracted Strategy: loads matching guidance
- Slash Commands → Concentrate Forces: loads matching guidance
- Slash Commands → Spark Prairie Fire: loads matching guidance
- Slash Commands → Overall Planning: loads matching guidance
- Workflow Compositions → Contradiction Analysis: coordinates method use
- Workflow Compositions → Practice Cognition: coordinates method use
- Workflow Compositions → Investigation First: coordinates method use
- Workflow Compositions → Mass Line: coordinates method use
- Workflow Compositions → Self-Criticism: coordinates method use
- Workflow Compositions → Protracted Strategy: coordinates method use
- Workflow Compositions → Concentrate Forces: coordinates method use
- Workflow Compositions → Spark Prairie Fire: coordinates method use
- Workflow Compositions → Overall Planning: coordinates method use
- AI Agent → Subagents: delegates investigation or review
- User → CLI Entry: runs install or uninstall
- CLI Entry → Platform Detection: detects and resolves targets
- CLI Entry → Asset Installer: installs or removes assets
- Asset Installer → Agent Hosts: copies assets to host directories
- Agent Hosts → Session Hooks: runs session integration