SILVER/NODEDC_1C
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Усилить агентную валидацию phase12 replay

Browse Source
This commit is contained in:
dctouch
2026-06-02 10:20:20 +03:00
parent 7995980215
commit 5c3ef746c0
6 changed files with 756 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
llm_normalizer/data/autorun_generators/history.json
+88
View File
@@ -1,4 +1,92 @@
[
{
"generation_id": "gen-ag06020717-08d833",
"created_at": "2026-06-02T07:17:06+00:00",
"mode": "saved_user_sessions",
"title": "AGENT | Phase 12 wider multi-trajectory authority replay",
"count": 20,
"domain": "address_phase12_wider_saved_session_pool",
"questions": [
"приветик - че как там дела",
"Альтернатива Плюс",
"расскажи что можешь интересного",
"кайф - что там на складе по остаткам?",
"а исторические остатки на другие даты умеешь?",
"март 2016",
"По выбранному объекту \"Рабочая станция универсального специалиста (индивидуальное изготовление)\": где взяли это?",
"а кому продали?",
"ндс можешь прикинуть на дату покупки рабочей станции?",
"кто у нас самый доходный клиент за все время",
"кто нам должен денег на май 2017",
"а какой ндс мы должны примерно заплатить за этот период?",
"мы должны комуто денег на сегодня?",
"а нам?",
"ты умеешь считать дельту по договорам?",
"по чепурнову покажи все доки",
"а по свк",
"хвосты покажи по счету 60 на август 2022",
"Есть ли остатки товара, которые закупались очень давно",
"а по Альтернативе Плюс сколько лет активности в базе 1С?"
],
"generated_by": "codex_agent",
"saved_case_set_file": "assistant_autogen_saved_user_sessions_20260602071706_gen-ag06020717-08d833.json",
"context": {
"llm_provider": null,
"model": null,
"assistant_prompt_version": null,
"decomposition_prompt_version": null,
"prompt_fingerprint": null,
"autogen_personality_id": null,
"autogen_personality_prompt": null,
"source_session_id": null,
"saved_session_file": "assistant_saved_session_20260602071706_gen-ag06020717-08d833.json",
"saved_case_set_kind": "agent_semantic_scenario",
"agent_run": true,
"agent_focus": "Phase12 wide replay: organization authority, inventory selected-object, VAT/date carryover, payables/receivables today reset, counterparty docs, old stock and activity age.",
"architecture_phase": "turnaround_11",
"source_spec_file": "X:\\1C\\NDC_1C\\docs\\orchestration\\address_truth_harness_phase12_wider_saved_session_pool.json",
"scenario_id": "address_truth_harness_phase12_wider_saved_session_pool",
"semantic_tags": [
"aggregate_revenue",
"bridge_inventory_to_vat",
"capability_over_followup",
"company_authority",
"company_selected",
"counterparty_followup",
"counterparty_root",
"cross_domain_pivot",
"date_carryover",
"display_label_integrity",
"documents",
"historical_date_anchor",
"human_answer_quality",
"inventory_aging",
"inventory_provenance",
"inventory_root",
"inventory_sale_trace",
"late_session_stability",
"meta_capability",
"meta_historical_capability",
"meta_smalltalk",
"organization_activity_age",
"organization_authority",
"proactive_scope_offer",
"same_date_restore",
"same_period_restore",
"selected_object",
"settlements_account_60",
"settlements_mirror_followup",
"settlements_payables",
"settlements_receivables",
"tail_authority_proof",
"today_scope",
"vat_followup"
],
"validation_status": "accepted_domain_case_loop_scenario",
"validated_run_dir": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"saved_after_validated_replay": true
}
},
{
"generation_id": "gen-ag06020614-947fd1",
"created_at": "2026-06-02T06:14:41+00:00",
llm_normalizer/data/autorun_generators/saved_sessions/assistant_saved_session_20260602071706_gen-ag06020717-08d833.json
+363
View File
@@ -0,0 +1,363 @@
{
"saved_at": "2026-06-02T07:17:06+00:00",
"generation_id": "gen-ag06020717-08d833",
"mode": "saved_user_sessions",
"title": "AGENT | Phase 12 wider multi-trajectory authority replay",
"agent_run": true,
"questions": [
"приветик - че как там дела",
"Альтернатива Плюс",
"расскажи что можешь интересного",
"кайф - что там на складе по остаткам?",
"а исторические остатки на другие даты умеешь?",
"март 2016",
"По выбранному объекту \"Рабочая станция универсального специалиста (индивидуальное изготовление)\": где взяли это?",
"а кому продали?",
"ндс можешь прикинуть на дату покупки рабочей станции?",
"кто у нас самый доходный клиент за все время",
"кто нам должен денег на май 2017",
"а какой ндс мы должны примерно заплатить за этот период?",
"мы должны комуто денег на сегодня?",
"а нам?",
"ты умеешь считать дельту по договорам?",
"по чепурнову покажи все доки",
"а по свк",
"хвосты покажи по счету 60 на август 2022",
"Есть ли остатки товара, которые закупались очень давно",
"а по Альтернативе Плюс сколько лет активности в базе 1С?"
],
"metadata": {
"assistant_prompt_version": null,
"decomposition_prompt_version": null,
"prompt_fingerprint": null,
"agent_focus": "Phase12 wide replay: organization authority, inventory selected-object, VAT/date carryover, payables/receivables today reset, counterparty docs, old stock and activity age.",
"architecture_phase": "turnaround_11",
"source_spec_file": "X:\\1C\\NDC_1C\\docs\\orchestration\\address_truth_harness_phase12_wider_saved_session_pool.json",
"scenario_id": "address_truth_harness_phase12_wider_saved_session_pool",
"semantic_tags": [
"aggregate_revenue",
"bridge_inventory_to_vat",
"capability_over_followup",
"company_authority",
"company_selected",
"counterparty_followup",
"counterparty_root",
"cross_domain_pivot",
"date_carryover",
"display_label_integrity",
"documents",
"historical_date_anchor",
"human_answer_quality",
"inventory_aging",
"inventory_provenance",
"inventory_root",
"inventory_sale_trace",
"late_session_stability",
"meta_capability",
"meta_historical_capability",
"meta_smalltalk",
"organization_activity_age",
"organization_authority",
"proactive_scope_offer",
"same_date_restore",
"same_period_restore",
"selected_object",
"settlements_account_60",
"settlements_mirror_followup",
"settlements_payables",
"settlements_receivables",
"tail_authority_proof",
"today_scope",
"vat_followup"
],
"validation_status": "accepted_domain_case_loop_scenario",
"validated_run_dir": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"saved_after_validated_replay": true,
"save_gate": {
"schema_version": "agent_semantic_save_gate_v1",
"validation_status": "accepted_domain_case_loop_scenario",
"validated_run_dir": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"final_status": "accepted",
"execution_status": "partial",
"scenario_id": "address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"steps_total": 20,
"effective_runtime": {
"manifest_path": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10\\effective_runtime.json",
"runner": "domain_case_loop.run-scenario",
"git_sha": "7995980215b6a9840c42c1c7ef73d6faa6612c37",
"backend_url": "http://127.0.0.1:8787",
"mcp_proxy_url": "http://127.0.0.1:6003",
"llm_provider": "local",
"llm_model": "unsloth/qwen3-30b-a3b-instruct-2507",
"temperature": 0.8,
"max_output_tokens": 900,
"prompt_version": "normalizer_v2_0_2",
"prompt_source": "file",
"prompt_hash": "f36e36a5e491cd24511b380e0c0059cb01aea5e2af738fb9f01671cd291735c1",
"prompt_registry_status": "pass"
},
"saved_after_validated_replay": true
}
},
"source_session_id": null,
"session": {
"session_id": null,
"mode": "agent_semantic_run",
"items": [
{
"message_id": "agent-user-001",
"role": "user",
"text": "приветик - че как там дела",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-002",
"role": "user",
"text": "Альтернатива Плюс",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-003",
"role": "user",
"text": "расскажи что можешь интересного",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-004",
"role": "user",
"text": "кайф - что там на складе по остаткам?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-005",
"role": "user",
"text": "а исторические остатки на другие даты умеешь?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-006",
"role": "user",
"text": "март 2016",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-007",
"role": "user",
"text": "По выбранному объекту \"Рабочая станция универсального специалиста (индивидуальное изготовление)\": где взяли это?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-008",
"role": "user",
"text": "а кому продали?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-009",
"role": "user",
"text": "ндс можешь прикинуть на дату покупки рабочей станции?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-010",
"role": "user",
"text": "кто у нас самый доходный клиент за все время",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-011",
"role": "user",
"text": "кто нам должен денег на май 2017",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-012",
"role": "user",
"text": "а какой ндс мы должны примерно заплатить за этот период?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-013",
"role": "user",
"text": "мы должны комуто денег на сегодня?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-014",
"role": "user",
"text": "а нам?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-015",
"role": "user",
"text": "ты умеешь считать дельту по договорам?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-016",
"role": "user",
"text": "по чепурнову покажи все доки",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-017",
"role": "user",
"text": "а по свк",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-018",
"role": "user",
"text": "хвосты покажи по счету 60 на август 2022",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-019",
"role": "user",
"text": "Есть ли остатки товара, которые закупались очень давно",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
},
{
"message_id": "agent-user-020",
"role": "user",
"text": "а по Альтернативе Плюс сколько лет активности в базе 1С?",
"created_at": "2026-06-02T07:17:06+00:00",
"reply_type": null,
"trace_id": null,
"debug": null
}
],
"agent_run": true,
"metadata": {
"assistant_prompt_version": null,
"decomposition_prompt_version": null,
"prompt_fingerprint": null,
"agent_focus": "Phase12 wide replay: organization authority, inventory selected-object, VAT/date carryover, payables/receivables today reset, counterparty docs, old stock and activity age.",
"architecture_phase": "turnaround_11",
"source_spec_file": "X:\\1C\\NDC_1C\\docs\\orchestration\\address_truth_harness_phase12_wider_saved_session_pool.json",
"scenario_id": "address_truth_harness_phase12_wider_saved_session_pool",
"semantic_tags": [
"aggregate_revenue",
"bridge_inventory_to_vat",
"capability_over_followup",
"company_authority",
"company_selected",
"counterparty_followup",
"counterparty_root",
"cross_domain_pivot",
"date_carryover",
"display_label_integrity",
"documents",
"historical_date_anchor",
"human_answer_quality",
"inventory_aging",
"inventory_provenance",
"inventory_root",
"inventory_sale_trace",
"late_session_stability",
"meta_capability",
"meta_historical_capability",
"meta_smalltalk",
"organization_activity_age",
"organization_authority",
"proactive_scope_offer",
"same_date_restore",
"same_period_restore",
"selected_object",
"settlements_account_60",
"settlements_mirror_followup",
"settlements_payables",
"settlements_receivables",
"tail_authority_proof",
"today_scope",
"vat_followup"
],
"validation_status": "accepted_domain_case_loop_scenario",
"validated_run_dir": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"saved_after_validated_replay": true,
"save_gate": {
"schema_version": "agent_semantic_save_gate_v1",
"validation_status": "accepted_domain_case_loop_scenario",
"validated_run_dir": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"final_status": "accepted",
"execution_status": "partial",
"scenario_id": "address_truth_harness_phase12_wider_saved_session_pool_20260602_p10",
"steps_total": 20,
"effective_runtime": {
"manifest_path": "artifacts\\domain_runs\\address_truth_harness_phase12_wider_saved_session_pool_20260602_p10\\effective_runtime.json",
"runner": "domain_case_loop.run-scenario",
"git_sha": "7995980215b6a9840c42c1c7ef73d6faa6612c37",
"backend_url": "http://127.0.0.1:8787",
"mcp_proxy_url": "http://127.0.0.1:6003",
"llm_provider": "local",
"llm_model": "unsloth/qwen3-30b-a3b-instruct-2507",
"temperature": 0.8,
"max_output_tokens": 900,
"prompt_version": "normalizer_v2_0_2",
"prompt_source": "file",
"prompt_hash": "f36e36a5e491cd24511b380e0c0059cb01aea5e2af738fb9f01671cd291735c1",
"prompt_registry_status": "pass"
},
"saved_after_validated_replay": true
}
}
}
}
llm_normalizer/data/eval_cases/assistant_autogen_saved_user_sessions_20260602071706_gen-ag06020717-08d833.json
+85
View File
@@ -0,0 +1,85 @@
{
"suite_id": "assistant_saved_session_gen-ag06020717-08d833",
"suite_version": "0.1.0",
"schema_version": "assistant_saved_session_suite_v0_1",
"generated_at": "2026-06-02T07:17:06+00:00",
"generation_id": "gen-ag06020717-08d833",
"mode": "saved_user_sessions",
"title": "AGENT | Phase 12 wider multi-trajectory authority replay",
"domain": "address_phase12_wider_saved_session_pool",
"scenario_count": 1,
"case_ids": [
"SAVED-001"
],
"cases": [
{
"case_id": "SAVED-001",
"scenario_tag": "agent_saved_user_sessions",
"title": "AGENT | Phase 12 wider multi-trajectory authority replay",
"question_type": "followup",
"broadness_level": "medium",
"turns": [
{
"user_message": "приветик - че как там дела"
},
{
"user_message": "Альтернатива Плюс"
},
{
"user_message": "расскажи что можешь интересного"
},
{
"user_message": "кайф - что там на складе по остаткам?"
},
{
"user_message": "а исторические остатки на другие даты умеешь?"
},
{
"user_message": "март 2016"
},
{
"user_message": "По выбранному объекту \"Рабочая станция универсального специалиста (индивидуальное изготовление)\": где взяли это?"
},
{
"user_message": "а кому продали?"
},
{
"user_message": "ндс можешь прикинуть на дату покупки рабочей станции?"
},
{
"user_message": "кто у нас самый доходный клиент за все время"
},
{
"user_message": "кто нам должен денег на май 2017"
},
{
"user_message": "а какой ндс мы должны примерно заплатить за этот период?"
},
{
"user_message": "мы должны комуто денег на сегодня?"
},
{
"user_message": "а нам?"
},
{
"user_message": "ты умеешь считать дельту по договорам?"
},
{
"user_message": "по чепурнову покажи все доки"
},
{
"user_message": "а по свк"
},
{
"user_message": "хвосты покажи по счету 60 на август 2022"
},
{
"user_message": "Есть ли остатки товара, которые закупались очень давно"
},
{
"user_message": "а по Альтернативе Плюс сколько лет активности в базе 1С?"
}
]
}
]
}