Opportunity Radar Agent Skill

Use the repository root SKILL.md as the canonical operating procedure. This file is a portable skill package entrypoint for agent environments that discover skills under skills/<name>/SKILL.md.

Required behavior

Follow this causal chain:

Need → Change → Career-to-Founder Fit → Signals → Payment → Industry Formation → Capital Heat → Economic Quality → Power → Operator Fit → Evidence Confidence → Why-Not-Yet → Window → Paid Pilot → Recurrence → Assetization → Real Option → Decision Ledger

Always:

• distinguish fact, inference and hypothesis;
• state geography and as-of date for current analysis;
• separate User / Beneficiary / Buyer / Payer;
• treat VC/angel/strategic funding as a capital signal, not customer proof;
• report Capital Heat separately as C0 Cold / C1 Emerging / C2 Building / C3 Hot / C4 Euphoric / C5 Unwinding;
• run Capital Heat × Payment × Retention × Contribution Economics × Competition × Power Pool before the verdict;
• do not return START for C4 Euphoric when payment and retention are weak or materially unverified;
• report Signal Score, Mobility Score, Evidence Coverage and Confidence separately;
• actively search for the strongest hostile explanation;
• require a 30–60 day paid test before large commitment when feasible;
• define leading, confirmation and kill criteria;
• identify the Power Pool and an assetization path;
• preserve dated decisions rather than rewriting history;
• use failure / false-positive backtests when the category is hype-sensitive;
• treat the static demo/ as a heuristic front-end only, never as live market evidence or a success-probability calculator.

Public artifact rule

For public README, examples, datasets, benchmarks or demos, use synthetic / fictional operator archetypes only. Never expose private user context or infer personal facts into a public artifact.

Reference loading

Load only what the task needs:

• references/capital-heat-lens.md for VC / angel / strategic capital;
• references/hype-divergence.md for capital-vs-customer divergence;
• cases/failure-backtests/matrix.md for historical anti-hype patterns;
• references/evidence-confidence.md for evidence coverage and confidence;
• references/window-and-why-not-yet.md for timing and hostile explanations;
• references/career-to-founder-transition.md for employment-to-ownership mobility;
• cases/failure-backtests/ for historical false positives;
• references/output-template.md for the final output structure.

Verdicts

Return exactly one primary verdict:

START | BUY A REAL OPTION | WATCH | REJECT

Never interpret a score as a probability of success.