How 20240610ldx-hub/mao-chiang-wisdom-skills works
The repository publishes two installable historical-reasoning Skills and a reusable Debate Arena. The main user workflow is either to install a Skill and use its framework in Codex, or run the Arena with two Skill files to generate a staged debate transcript, optionally adding fact-checking and a judge report. The graph also shows the Skills’ supporting materials and selected published evaluations and debate artifacts. Arena control flow is based on sampled source; broader Skill behavior is described by the README rather than runtime code.
Main entry points: orchestrator.py Debate runner · SKILL.md Mao Skill · cli.py CLI entry · loader.py Skill loader
Components 18
Reasoning Skills
- Mao Skill
skills/mao-zedong-wisdom/SKILL.md - Mao principles
skills/mao-zedong-wisdom/assets - Evidence and voice
skills/mao-zedong-wisdom/references - Chiang Skill
skills/chiang-kai-shek-wisdom/SKILL.md
Debate Arena
- CLI entry
debate_arena/cli.py - Runtime config
debate_arena/config.py - Debate runner
debate_arena/orchestrator.py - Skill loader
debate_arena/loader.py - Prompt rendering
debate_arena/prompting.py - Debate prompts
debate_arena/prompts - Debate records
debate_arena/models.py - LLM provider
debate_arena/llm_client.py - Markdown renderer
debate_arena/render.py
Published Records
- Debate artifacts
debates/mao-vs-chiang - Skill evaluations
evaluations
Other components
- Skill user
- LLM service
- Debate transcript
Connections 19
- Skill user → Mao Skill: uses
- Skill user → Chiang Skill: uses
- Mao Skill → Mao principles: references as needed
- Mao Skill → Evidence and voice: references as needed
- Skill user → CLI entry: runs
- CLI entry → Runtime config: builds config
- CLI entry → Debate runner: starts debate
- Debate runner → Skill loader: loads Skills
- Skill loader → Mao Skill: reads input
- Skill loader → Chiang Skill: reads input
- Debate runner → Prompt rendering: renders prompts
- Prompt rendering → Debate prompts: loads templates
- Debate runner → LLM provider: requests completions
- LLM provider → LLM service: sends requests
- Debate runner → Debate records: builds transcript
- Debate runner → Markdown renderer: renders transcript
- Markdown renderer → Debate transcript: produces Markdown
- Skill user → Skill evaluations: reviews
- Skill user → Debate artifacts: reviews