
Judgment Training (Naval)
Guides the agent to reason from first principles and apply mental models for direction and credibility questions
- Repository
- kangarooking/cangjie-skill
- Stars
- 11,998
- Licence
- MIT
- Files
- 3 files
What it does
This is a reading-based thinking guide, distilled from Naval Ravikant's book. Once installed, when you ask the agent to help judge a long-term direction, assess whether an expert's claim is credible, or improve your decision-making, it works through the problem by checking whether the reasoning can be rebuilt from basic principles and by applying mental models such as compounding, principal-agent problems, falsifiability and black swans. It also flags claims that sound impressive but cannot be derived from fundamentals.
Great for
- People facing a big strategic or life decision who want a structured way to think it through
- Anyone evaluating whether a claim or expert is genuinely knowledgeable
- Students or learners building judgment in an unfamiliar field
Skip it if
- Quick, concrete execution tasks: the skill itself says to act first and reflect later, and it adds unnecessary deliberation (reason: it is meant for direction-setting, not step-by-step work)
- Simple factual lookups, where a full first-principles derivation is unnecessary
Try asking
- “Help me judge whether it's worth switching careers into data science”
- “Is this expert's claim about intermittent fasting credible? Break down what they actually know”
- “I want to build better judgment about investing. Which mental models should I learn first?”
- “Check whether my reasoning for launching this product holds up from first principles”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 3 original files, the MIT licence and an attribution note.
judgment-training.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip judgment-training.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip judgment-training.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- State the decision and its time horizon clearly so the agent can judge whether consequences are long-term
- Ask the agent to name which mental models it applied and where they may not fit your situation
- Pair it with related skills it mentions (decision-heuristics, principal-agent) for more concrete rules
What's inside
- SKILL.md
- test-prompts.json
- test-results.md
Redistributed unmodified from kangarooking/cangjie-skill under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.