<2–4 sentence statement of the rule. State it as a hard rule, not a suggestion. Name the specific failure mode it prevents.> **Why it matters**: <Why violating this rule produces a wrong analysis. Include the mechanism — what specific score / target / recommendation gets corrupted.> **Concrete failure** (optional but preferred): <One specific dated case where this pitfall produced a wrong answer. Format: ticker, date, what was claimed, what actually happened, magnitude of error.> **How to apply**: 1. <Numbered steps, each one a concrete check or action.> 2. <Make them executable — what to compute, what threshold to compare against, what file to load.> 3. <Cite the script / agent / stage that should enforce the check.> **When the rule does NOT apply**: - <Edge case 1 — explicit exception with reason> - <Edge case 2> **Cross-references**: - Pitfall NN — <related pitfall, why related> - `references/<file>.md` — <related framework> - Stage NN (`agents/<agent>.md`)

--- title: severity: HIGH | MEDIUM | LOW appliesTo: tags: --- ## <2–4 sentence statement of the rule. State it as a hard rule, not a suggestion. Name the specific failure mode it prevents.> **Why it matters**: <Why violating this rule produces a wrong analysis. Include the mechanism — what specific score / target / recommendation gets corrupted.> **Concrete failure** (optional but preferred): <One specific dated case where this pitfall produced a wrong answer. Format: ticker, date, what was claimed, what actually happened, magnitude of error.> **How to apply**: 1. <Numbered steps, each one a concrete check or action.> 2. <Make them executable — what to compute, what threshold to compare against, what file to load.> 3. <Cite the script / agent / stage that should enforce the check.> **When the rule does NOT apply**: - <Edge case 1 — explicit exception with reason> - <Edge case 2> **Cross-references**: - Pitfall NN — <related pitfall, why related> - `references/<file>.md` — <related framework> - Stage NN (`agents/<agent>.md`) — <enforcement point>