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

Запустить Open-World Breadth через бизнес-обзор компании

Browse Source
This commit is contained in:
dctouch
2026-05-02 00:09:28 +03:00
parent 6df2018086
commit 284b201912
9 changed files with 477 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/ARCH/11 - architecture_turnaround/21 - current_status_canon_2026-05-01.md
+7 -5
View File
@@ -13,7 +13,8 @@ If another document says `78%`, `87%`, `92%`, or `85%` for a module that is now
- `Post-F Semantic Integrity Hardening`: `99%`, operationally closed as a hardening slice and now used as a regression gate.
- `Inventory Stock Open-World Breadth Proof`: `100%` for the declared inventory-stock scenario pack, not for arbitrary inventory questions.
- `Planner Autonomy Consolidation`: `100%` for the declared phase83 planner-brain slice, including catalog alignment, live-readiness gating, checked-source sanitation, and accepted mixed replay.
- Active next pressure: broader `Open-World Bounded Autonomy Breadth` over unfamiliar 1C asks, while keeping Post-F and phase83 as regression gates.
- Active next module: broader `Open-World Bounded Autonomy Breadth` over unfamiliar 1C asks, while keeping Post-F and phase83 as regression gates.
- First active slice: `Business Overview Evidence Fusion`, tracked in `22 - open_world_bounded_autonomy_breadth_2026-05-01.md`.
## Reporting Rule
@@ -70,9 +71,10 @@ For current planning, read:
1. `README.md`
2. this document
3. `20 - planner_autonomy_consolidation_2026-05-01.md`
4. `19 - inventory_stock_open_world_breadth_proof_2026-05-01.md`
5. `17 - post_f_semantic_integrity_hardening_2026-04-23.md`
6. `16 - data_need_graph_and_open_world_mcp_plan_2026-04-22.md`
3. `22 - open_world_bounded_autonomy_breadth_2026-05-01.md`
4. `20 - planner_autonomy_consolidation_2026-05-01.md`
5. `19 - inventory_stock_open_world_breadth_proof_2026-05-01.md`
6. `17 - post_f_semantic_integrity_hardening_2026-04-23.md`
7. `16 - data_need_graph_and_open_world_mcp_plan_2026-04-22.md`
Documents `01` through `15` remain valuable, but mostly as the historical architecture trail.
docs/ARCH/11 - architecture_turnaround/22 - open_world_bounded_autonomy_breadth_2026-05-01.md
+101
View File
@@ -0,0 +1,101 @@
# 22 - Open-World Bounded Autonomy Breadth (2026-05-01)
## Purpose
This note opens the next active module after:
- Post-F Semantic Integrity Hardening operational closure;
- inventory-stock bounded breadth proof;
- Planner Autonomy Consolidation phase83 closure.
The goal is not to reopen those slices.
The goal is to grow the assistant from reviewed route families toward broader bounded 1C autonomy:
`user business ask -> data-need graph -> catalog route fabric -> reviewed MCP evidence -> truth gate -> useful analyst answer`
## Entry Baseline
Already closed and kept as regression gates:
- Post-F semantic integrity: explicit current-turn meaning beats stale scope, stale focus object, and wrong post-pivot arbitration.
- Planner phase83: selected chain must align with reviewed catalog templates and replay artifacts must expose that alignment.
- Inventory breadth proof: a domain scenario tree can be accepted only through user-facing business correctness, not route labels alone.
The current active pressure is:
- unfamiliar or broad human 1C asks still need more bounded breadth;
- broad business questions must become evidence-guided analyst answers, not generic chat summaries;
- new capabilities must grow through reusable data-need and catalog surfaces, not one-off prompt patches.
## Slice 1 - Business Overview Evidence Fusion
User-facing trigger examples:
- `Как ты оценишь деятельность компании?`
- `Дай полный анализ компании`
- `Сделай LLM-аудит бизнеса`
- `Что думаешь о компании в целом?`
### Problem
Before this slice, broad business evaluation was protected from stale lifecycle replay, but it could still be too thin:
- it summarized context, but did not consistently surface confirmed metrics;
- it did not clearly separate cash-flow signals from profit/margin claims;
- it could under-answer when the user expected a mature analyst-style company overview.
### Current Implementation Boundary
The first implementation step is intentionally bounded:
- it does not invent fresh facts;
- it fuses already-confirmed MCP/session evidence from recent grounded answers;
- it recognizes broader company-analysis wording as `broad_business_evaluation`;
- it extracts useful analyst signals from confirmed value-flow, bidirectional net-flow, ranking, lifecycle, and inventory evidence;
- it states that profit, margin, debt quality, and full company health remain unproved unless corresponding evidence exists.
This is not yet the final automatic multi-probe company-analysis chain.
It is the safe first slice: better user-facing business overview without bypassing evidence gates.
### Runtime Contract
When this bridge answers, it must:
- start with a direct business summary;
- list confirmed metrics separately from interpretation;
- call out cash-flow direction and net spread only as cash evidence, not as profit;
- mention top counterparty/customer only when a ranked value-flow evidence slice exists;
- keep profit/margin/debt/VAT as explicit missing evidence when not checked;
- avoid raw MCP, planner, catalog, route, primitive, or debug wording.
### Next Slice
Promote this bridge into a real planner route:
- add a reviewed `business_overview` catalog chain template;
- let the data-need graph model broad evaluation as a composed evidence need;
- run bounded fresh probes for year turnover, top customers, incoming/outgoing/net flow, debt, VAT, and inventory context where available;
- return a layered analyst answer with exact evidence, bounded inference, unknowns, and recommended next probes.
## Acceptance Signals
The slice is healthy when:
- broad analysis wording lands on the business-overview contour;
- the answer is materially more informative than a generic recap;
- confirmed metrics are visibly separated from LLM-style interpretation;
- profit/margin are not claimed without supporting evidence;
- Post-F stale-scope and phase83 catalog-alignment canaries remain green.
## Validation
Initial local validation:
- `npm.cmd test -- assistantTurnMeaningPolicy.test.ts assistantLivingChatRuntimeAdapter.test.ts`: passed `20/20`.
- `npm.cmd test -- assistantTurnMeaningPolicy.test.ts assistantLivingChatRuntimeAdapter.test.ts assistantRoutePolicy.test.ts assistantMcpDiscoveryResponsePolicy.test.ts`: passed `56/56`.
- `npm.cmd run build`: passed.
- graphify rebuild: `5977 nodes`, `12983 edges`, `137 communities`.
Graphify must be rebuilt after this code/doc slice before commit.
docs/ARCH/11 - architecture_turnaround/README.md
+7 -2
View File
@@ -39,6 +39,7 @@ This package answers the next question:
19. [19 - inventory_stock_open_world_breadth_proof_2026-05-01.md](./19%20-%20inventory_stock_open_world_breadth_proof_2026-05-01.md)
20. [20 - planner_autonomy_consolidation_2026-05-01.md](./20%20-%20planner_autonomy_consolidation_2026-05-01.md)
21. [21 - current_status_canon_2026-05-01.md](./21%20-%20current_status_canon_2026-05-01.md)
22. [22 - open_world_bounded_autonomy_breadth_2026-05-01.md](./22%20-%20open_world_bounded_autonomy_breadth_2026-05-01.md)
## Current Status Snapshot (2026-05-01)
@@ -48,7 +49,8 @@ Status canon for planning:
- Post-F Semantic Integrity Hardening is operationally closed at `99%` and should now be used as a regression gate, not as the active module denominator.
- Planner Autonomy Consolidation is closed at `100%` for the declared phase83 planner-brain slice.
- The active next pressure is broader `Open-World Bounded Autonomy Breadth` over unfamiliar 1C asks, with Post-F and phase83 retained as semantic canaries.
- The active next module is now `Open-World Bounded Autonomy Breadth` over unfamiliar 1C asks, with Post-F and phase83 retained as semantic canaries.
- The first active slice is `Business Overview Evidence Fusion`: broad company-analysis wording now produces a richer evidence-grounded business overview from confirmed MCP/session facts instead of a thin generic summary.
- The short source of truth for status wording is [21 - current_status_canon_2026-05-01.md](./21%20-%20current_status_canon_2026-05-01.md).
It now documents a turnaround that is already operational in code, already materially past the acute regression breakpoint, and already moved through bounded MCP autonomy, Post-F hardening, inventory breadth proof, and the declared Planner Autonomy slice:
@@ -113,10 +115,11 @@ Current honest status:
- pre-multidomain readiness: `~90%`
- bounded-autonomy foundation readiness: `~89%`
- open-world bounded-autonomy readiness: `~85%`
- active Open-World Bounded Autonomy Breadth progress: `~12%`, with the first business-overview evidence-fusion slice started and locally tested; fresh multi-probe planner route is still pending
- Post-F semantic integrity module progress: `~99%` operationally closed, with remaining risk now treated as next-slice discovery rather than an open blocker inside the closed slice
- active inventory-stock breadth slice progress: `100%` for the declared scenario pack, not for arbitrary inventory questions
- Planner Autonomy Consolidation progress: `100%` for the declared module, with catalog-fabric, value-flow arbitration, lifecycle bounded inference, broad-evaluation bridge, inventory catalog templates, inventory runtime-boundary honesty, exact inventory recipe bridging, unambiguous metadata-surface lane inference, catalog chain-template scoring, structured chain-match contract exposure, runtime/debug propagation, subject-aware bidirectional comparison arbitration, structured catalog-alignment verdicts, representative alignment regression guard, catalog-alignment reason-code telemetry, explicit `alignment_status` propagation, truth-harness/acceptance-matrix surfacing, soft divergence warning, `catalog_alignment_ok` acceptance invariant, step-level expected catalog-alignment assertions, phase66 and phase32 spec alignment expectations, AGENT source-catalog surfacing, generated phase83 mixed planner-brain replay spec, checked-source user-facing error sanitation, surface-grounded catalog promotion, and guarded live phase83 acceptance validated. Broader unfamiliar 1C asks are now next-module breadth work rather than an open blocker inside this declared slice
- graph snapshot after latest rebuild: `5974 nodes`, `12974 edges`, `138 communities`
- graph snapshot after latest rebuild: `5977 nodes`, `12983 edges`, `137 communities`
- current regression-gate breakpoint:
- the validated hot paths are no longer structurally broken;
- flagship continuity collapse is no longer the primary risk;
@@ -205,6 +208,7 @@ For the detailed audit, current percentages, and remaining debt, read:
- [19 - inventory_stock_open_world_breadth_proof_2026-05-01.md](./19%20-%20inventory_stock_open_world_breadth_proof_2026-05-01.md)
- [20 - planner_autonomy_consolidation_2026-05-01.md](./20%20-%20planner_autonomy_consolidation_2026-05-01.md)
- [21 - current_status_canon_2026-05-01.md](./21%20-%20current_status_canon_2026-05-01.md)
- [22 - open_world_bounded_autonomy_breadth_2026-05-01.md](./22%20-%20open_world_bounded_autonomy_breadth_2026-05-01.md)
## Architectural Objects Of Planning
@@ -244,6 +248,7 @@ Read in this order:
20. `19 - inventory_stock_open_world_breadth_proof_2026-05-01.md`
21. `20 - planner_autonomy_consolidation_2026-05-01.md`
22. `21 - current_status_canon_2026-05-01.md`
23. `22 - open_world_bounded_autonomy_breadth_2026-05-01.md`
## Planning Rules