Исправить компактные follow-up ответы по денежному итогу 1С

This commit is contained in:
2026-05-23 17:30:01 +03:00
parent 91a9dfa3a7
commit f7314b0212
18 changed files with 876 additions and 16 deletions
@@ -0,0 +1,141 @@
{
"schema_version": "domain_truth_harness_spec_v1",
"scenario_id": "agent_cashflow_compact_display_variants_20260523",
"domain": "autonomy_business_answer_contract",
"title": "AGENT | Cashflow compact display variants",
"description": "Targeted AGENT replay for compact cashflow display modifiers: no counterparties, no breakdown, one-line and only-total follow-ups must preserve the 2020 company cashflow context.",
"bindings": {},
"steps": [
{
"step_id": "step_01_explicit_overview_anchor",
"title": "Create 2020 business overview context",
"question": "Теперь дай взрослый обзор за 2020 по компании: входящие, исходящие, нетто, топы, но банк в топах отдельно объясни как финансовый поток.",
"allowed_reply_types": ["partial_coverage", "factual_with_explanation", "factual"],
"required_answer_patterns_all": [
"47[\\s.]*628[\\s.]*853",
"43[\\s.]*763[\\s.]*351",
"3[\\s.]*865[\\s.]*501",
"12[\\s.]*792[\\s.]*194",
"12[\\s.]*093[\\s.]*465"
],
"forbidden_answer_patterns": ["runtime_", "planner_", "query_movements", "primitive"],
"criticality": "high",
"semantic_tags": ["business_overview", "context_anchor"]
},
{
"step_id": "step_02_no_counterparties",
"title": "No-counterparties means no breakdown, not exclusion",
"question": "сколько заработали деньгами без контрагентов?",
"allowed_reply_types": ["partial_coverage", "factual_with_explanation", "factual"],
"required_answer_patterns_all": [
"2020",
"47[\\s.]*628[\\s.]*853",
"43[\\s.]*763[\\s.]*351",
"3[\\s.]*865[\\s.]*501"
],
"forbidden_answer_patterns": [
"2026-05-23",
"получили\\s+0\\s*руб",
"нетто\\s+0\\s*руб",
"с исключением",
"исключением крупнейших",
"Комитет государственных услуг",
"Группа СВК",
"СБЕРБАНК",
"Что проверить дальше",
"runtime_",
"planner_",
"query_movements",
"primitive"
],
"criticality": "critical",
"semantic_tags": ["no_counterparty_breakdown", "display_modifier", "temporal_carryover"]
},
{
"step_id": "step_03_only_total",
"title": "Only total keeps compact cashflow",
"question": "только итог: пришло, ушло, нетто",
"allowed_reply_types": ["partial_coverage", "factual_with_explanation", "factual"],
"required_answer_patterns_all": [
"2020",
"47[\\s.]*628[\\s.]*853",
"43[\\s.]*763[\\s.]*351",
"3[\\s.]*865[\\s.]*501"
],
"forbidden_answer_patterns": [
"2026-05-23",
"получили\\s+0\\s*руб",
"нетто\\s+0\\s*руб",
"Комитет государственных услуг",
"Группа СВК",
"СБЕРБАНК",
"Что проверить дальше",
"runtime_",
"planner_",
"query_movements",
"primitive"
],
"criticality": "critical",
"semantic_tags": ["only_total", "display_modifier", "temporal_carryover"]
},
{
"step_id": "step_04_one_line_no_breakdown",
"title": "One-line no-breakdown stays cashflow",
"question": "дай одной строкой деньги без разбивки",
"allowed_reply_types": ["partial_coverage", "factual_with_explanation", "factual"],
"required_answer_patterns_all": [
"2020",
"47[\\s.]*628[\\s.]*853",
"43[\\s.]*763[\\s.]*351",
"3[\\s.]*865[\\s.]*501"
],
"forbidden_answer_patterns": [
"2026-05-23",
"получили\\s+0\\s*руб",
"нетто\\s+0\\s*руб",
"Комитет государственных услуг",
"Группа СВК",
"СБЕРБАНК",
"Что проверить дальше",
"runtime_",
"planner_",
"query_movements",
"primitive"
],
"criticality": "critical",
"semantic_tags": ["one_line", "no_breakdown", "display_modifier"]
},
{
"step_id": "step_05_no_details",
"title": "No-details wording remains direct totals",
"question": "сколько получили и заплатили без детализации",
"allowed_reply_types": ["partial_coverage", "factual_with_explanation", "factual"],
"required_answer_patterns_all": [
"2020",
"47[\\s.]*628[\\s.]*853",
"43[\\s.]*763[\\s.]*351",
"3[\\s.]*865[\\s.]*501"
],
"forbidden_answer_patterns": [
"2026-05-23",
"получили\\s+0\\s*руб",
"нетто\\s+0\\s*руб",
"Комитет государственных услуг",
"Группа СВК",
"СБЕРБАНК",
"Что проверить дальше",
"runtime_",
"planner_",
"query_movements",
"primitive"
],
"criticality": "critical",
"semantic_tags": ["no_details", "display_modifier", "direct_totals"]
}
],
"acceptance": {
"min_score": 80,
"max_unresolved_p0": 0,
"require_all_critical_steps_pass": true
}
}