stage-3-verify-templateslisted

# Stage 3 artifact templates ## error-analysis/summary.md (canonical R1 artifact — merged by the coordinator after all batch workers finish; `error-analysis/batch-NNN.md` worker files are frontmatter-exempt raw trace data) ```markdown --- artifact: error-analysis-summary stage: 3 gate: R1 status: draft evidence_grade: C rung: <rung — note judge substitute used> ... --- # Error analysis (open coding — ~100 traces to start; stop after ~20 consecutive traces with no new category) | # | input (ref) | output (ref) | error noted | category (taxonomy) | fix-now or eval? | |---|---|---|---|---|---| ## Failure taxonomy | Category | Count | Severity | Design answer | |---|---|---|---| ## Judge rung - Human review: user reviewed ___ outputs on YYYY-MM-DD / none - Model consensus: models used ___, agreement ___% ## Criteria (drifted freely while reading; final) ``` ## promise-scope.md ```markdown --- artifact: promise-scope ... gate: R1 ... status: draft # locked at R1 pass --- # Promise scope (locks at R1 — feeds positioning) - Customer acceptance threshold (from V1/V2, E-ids): ___ · signed into thresholds.r1_eval_pass_pct on YYYY-MM-DD - Eval result on real data: ___% - Failure style: visibly-wrong / plausibly-wrong → HITL required? ___ - Marginal cost per use: $___ vs V3 price $___ → margin: ___ - Hard constraints: ## The product PROMISES / REFUSES TO PROMISE ``` ## data-acquisition-plan.md (R1 OPEN only — one of the three OPEN deliverables) ```markdown --- artifact: data-acquis