Austrian Judgment Protocol

English | 简体中文

A production-grade judgment framework for founders, CEOs, and CTOs — six executable protocols from the Austrian school of economics, designed to be run by AI agents as a decision advisor.

Most business frameworks answer how to plan. This one answers the harder questions: who should decide, when, and on what evidence — before you commit capital, respond to a competitor, delegate a decision, or bet on a new market.

Who this is for

You are	Your decisions	What the protocol does
Founder	Is this opportunity real? Is the demand there? Should I commit capital now or stage it?	Demand-side three-stage test; capitalist's three judgments; uncertainty staging with kill criteria
CEO	Competitive response, strategy review, goal setting, org design	Diagnose before acting; kernel check; asymmetry audit; reject goal-fallacy framing
CTO / tech lead	Tech debt vs. feature velocity, delegation to teams/agents, architecture vs. business exploration	Judgment classification (original vs. derived); knowledge distribution; capability-boundary check
AI agent builder	Embedding judgment into agent workflows	Executable decision procedures + anti-patterns + traceable evidence coordinates — not free-text advice

What it solves

The situation	The common move (and the trap)	The protocol move
Sales declining	Attribute it to the market	Weak-signal check first (K04/KX03)
Competitor launches a free tier	Copy them immediately	Diagnose the real threat; respond from your own asymmetries (R01/R04)
New venture before committing money	Full commit, or endless "more data"	Three-stage demand test; stage the commitment, predefine kill criteria (P003, UX01)
Innovation stalled in a strong technical team	Hire more experts	Einstellung audit: one-sided knowledge actively reduces innovativeness (P006/P007)
"Who should decide this?"	By title, or by whoever shouts loudest	Original vs. derived judgment classification (FK02)
Head-office reports keep missing the real picture	More dashboards, more aggregation	Information anti-filtering: sample the raw periphery (HX01)

Why this framework

1. Not another planning framework. SWOT, OKR, and Porter answer "how do we plan?" They assume judgment is given. This framework makes judgment itself the object: who holds it, what evidence it rests on, how it fails.
2. Not a book summary. Theory is compressed into executable procedures and anti-patterns. All 84 claims are traceable to 9 primary sources (Kirzner, Mises, Hayek, Taleb, O'Driscoll & Rizzo, Rumelt, Foss & Klein, Packard, Christensen) with evidence tier marked — high = directly from the author's text, medium = framework inference.
3. A chain, not a grab bag. Six protocols form one judgment chain: opportunity discovery → demand validation → strategy diagnosis → knowledge distribution → judgment authority → uncertainty design. Using any one alone is incomplete by design.
4. Production-grade. Versioned (v1.0.0), machine-checkable claim ledger, cross-referenced dependencies, explicit scope and gaps per protocol.

How it works

The direct user is an AI agent (Hermes skill format): you describe a scenario, the agent routes to the relevant protocol and returns structured checks, anti-pattern audits, and evidence coordinates.

• Install: load SKILL.md as a skill; references/protocols/ holds the six protocols.
• Ask with a scenario: "analyze this competitive threat", "should we delegate this decision?", "evaluate this opportunity" → routing table in SKILL.md.
• Ask with a topic: alertness, case probability, strategy kernel, man on the spot, original judgment, three-stage process.

Quick start

# 1. Get the skill
git clone https://github.com/gootf/austrian-judgment-protocol.git
# 2. Copy SKILL.md + references/ into your agent's skills directory
#    (e.g. ~/.claude/skills/austrian-judgment-protocol/ or your agent's equivalent)

Then ask your agent — three ways it routes:

You say	What you get
"A competitor just launched a free tier. Analyze this threat."	strategy-diagnosis: symptom vs challenge, kernel check, asymmetry audit (R01/R04)
"Should we commit capital to this new venture now?"	consumer-sovereignty: three-stage demand test, staging + kill criteria (P003, UX01)
"Can we delegate this decision to the team?"	hayek-knowledge-problem: time-and-place? aggregable? delegate + coordinate (HX01)

What it deliberately does NOT do

• No prediction. Case probability is not computable (Mises); anyone selling you forecasts where the future is unique is selling a narrative fallacy (Taleb).
• No replacement of human judgment. Original judgment cannot be outsourced (Foss & Klein); the framework's own boundary is stated in its uncertainty protocol: AI handles puzzles, the mystery stays with the human decision-maker.
• No legal, financial, or investment advice.

Structure

SKILL.md        umbrella: routing table + usage
claims.yaml     84 claims, each traceable (author, book, evidence tier)
references/     protocols ×6 · glossary · patterns ×10 · cheatsheet

Sources

Kirzner · Mises · Taleb (×2) · O'Driscoll & Rizzo · Rumelt · Hayek · Foss & Klein · Packard · Christensen

License

MIT — see LICENSE.