docs: consolidate AES standards

.agents/skills/ai-os-orchestrator/SKILL.md

@@ -18,7 +18,7 @@ Before routing, verify and read:
 3. `PROJECT_CAPABILITIES.yaml` for capability locations;
 4. `.agents/skills/project-context/SKILL.md` for bounded context loading;
 5. `ChatGPT/[AI OS]/Knowledge/HANDOFF_PROTOCOL.md` and `HANDOFF_STYLE_STANDARD.md` when a cross-project handoff is required;
-6. `GOAL_MODE.md` and `AUTONOMOUS_EXECUTION_STANDARD.md` when execution, correction, validation, or terminal reporting is required.
+6. `GOAL_MODE.md` and `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` when execution, correction, validation, or terminal reporting is required.
 
 Treat those files as the semantic owners. Do not copy their routing tables, capability locations, project methodology, or merge policy into this skill.
 
@@ -74,7 +74,10 @@ pointer may cache only `execution_id` and `record_ref` after a behavioral test
 shows it is needed; it is never canonical state and must not duplicate the
 goal, acceptance criteria, requirements, defects, or authority state.
 
-If validation fails, follow the corrective-loop and authority rules in `AUTONOMOUS_EXECUTION_STANDARD.md`: register the defect, route it to its owner, make only an eligible minimal correction, rerun the same affected check, and keep `[Codex]`'s stricter one-correction limit. Never weaken acceptance criteria to terminate.
+If validation fails, follow the corrective-loop and authority rules in
+`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`: register the defect, route
+it to its owner, make only an eligible minimal correction, rerun the same
+affected check, and keep `[Codex]`'s stricter one-correction limit. Never weaken acceptance criteria to terminate.
 
 `Invoke AI-OS` does not expand authority. It must not auto-execute owner-frozen policy changes, merge, deploy, production promotion, destructive or low-reversibility actions, or actions requiring unavailable credentials, permissions, money, legal authority, or physical action.
 

.github/CODEOWNERS

@@ -18,8 +18,8 @@
 /ARCHIVE_MAP.md @sergstack
 /UPLOAD_GUIDE.md @sergstack
 /REPO_PATHS.md @sergstack
-/AUTONOMOUS_EXECUTION_STANDARD.md @sergstack
-/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md @sergstack
+/docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md @sergstack
+/docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md @sergstack
 /ROUTING_RULES.md @sergstack
 /HANDOFF_STYLE_STANDARD.md @sergstack
 /docs/standards/PROMPT_QA_FACTORY.md @sergstack

AGENTS.md

@@ -12,7 +12,7 @@ Atomic task packages remain available for advanced, high-risk, strict, or ultra-
 
 ## Autonomous Execution (AES)
 
-Execution follows [`AUTONOMOUS_EXECUTION_STANDARD.md`](AUTONOMOUS_EXECUTION_STANDARD.md)
+Execution follows [`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`](docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md)
 together with Goal Mode, the Codex autonomy policy, and applicable project
 extensions. For normal local and reversible work, infer the smallest safe
 scope, execute, validate, record assumptions, and continue without asking

ChatGPT/[AI OS]/Knowledge/AIOS_02_GOVERNANCE_AND_EVIDENCE_BUNDLE_SEMANTICS.md

@@ -9,13 +9,13 @@ Do not use this pattern to justify autonomous retrieval:
 ## Autonomous Execution Standard
 `[AI OS]` is the canonical owner of the Autonomous Execution Standard (AES).
 Execution across all projects now also follows the canonical loop defined in
-`AUTONOMOUS_EXECUTION_STANDARD.md` at the repo root: requirements -> execution
+`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` at the repo root: requirements -> execution
 -> validation -> defect registration -> corrective action -> affected-scope
 rerun -> revalidation -> scope acceptance -> final evidence. It does not
 replace Goal Mode, routing, autonomy policy, or the merge policy in
 `GOAL_MODE.md`; it connects them into one closed loop, and the stricter rule
 wins on any conflict. `[AI OS]` also owns the generic project-extension
-interface in `AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`, which a project
+interface in `docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`, which a project
 implements to add domain-specific defect subtypes, evidence, and acceptance
 scopes without restating the canonical state machine or schema.
 New AES v2 records require Closure Review: it rechecks original goal, scope,

ChatGPT/[AI OS]/Knowledge_Bundles/AIOS_02_GOVERNANCE_AND_EVIDENCE.md

@@ -8,9 +8,9 @@ Compact upload artifact for [AI OS] covering governance and evidence.
 
 - `ChatGPT/[AI OS]/Knowledge/GOVERNANCE_RULES.md`
 - `ChatGPT/[AI OS]/Knowledge/ANTI_PATTERNS.md`
-- `AUTONOMOUS_EXECUTION_STANDARD.md`
-- `AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`
-- `AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`
+- `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`
+- `docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`
+- `docs/standards/AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`
 - `ChatGPT/[AI OS]/Knowledge/AIOS_02_GOVERNANCE_AND_EVIDENCE_BUNDLE_SEMANTICS.md`
 
 ## Upload target
@@ -22,7 +22,7 @@ ChatGPT Project Sources / Knowledge for `[AI OS]`.
 - production_promotion: no, unless explicitly accepted elsewhere
 - bundle_type: generated compact upload artifact
 - source_of_truth: declared granular source files
-- source_fingerprint: sha256:6e319b114f9955208dbf4e733a9db5d4d35f7a3b493c0d7d94680b6c107b9ab9
+- source_fingerprint: sha256:9c0c5e7191c595f05ed8671a4534147fe15a475f4afa3fbefd48d499a1eacb36
 - generator: scripts/build_knowledge_bundles.py
 
 ---
@@ -179,13 +179,13 @@ autonomous retrieval
 - скрыта неопределённость;
 - нет web-проверки для текущих AI-релизов.
 
-## From: `AUTONOMOUS_EXECUTION_STANDARD.md`
+## From: `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`
 
 # Autonomous Execution Standard (AES) v2.0.0
 Status: normative package with scoped advisory semantic validation.
 Canonical owner: `[AI OS]`.
-Canonical source path: `AUTONOMOUS_EXECUTION_STANDARD.md` (this file, repo root).
-Companion contract: `AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`.
+Canonical source path: `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`.
+Companion contract: `docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`.
 Current schema: `schemas/autonomous_execution_record.schema.json`.
 Historical v1 schema: `schemas/autonomous_execution_record.v1.schema.json`.
 This document is the single canonical source for the Autonomous Execution
@@ -253,7 +253,7 @@ remain historical evidence and are not rewritten.
 3. approved task package
 4. project instructions
 5. applicable project-specific execution extension
-6. this document (AUTONOMOUS_EXECUTION_STANDARD.md)
+6. this document (`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`)
 7. supporting playbooks, templates and examples
 ```
 General rule: when two applicable rules conflict, the stricter constraint on
@@ -270,7 +270,7 @@ Consequences:
 ## 2. Canonical ownership
 ```yaml
 canonical_owner: "[AI OS]"
-canonical_source_path: "AUTONOMOUS_EXECUTION_STANDARD.md"
+canonical_source_path: "docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md"
 derived_artifacts: []   # none registered in Phase 1
 supersedes: []
 superseded_by: []
@@ -418,7 +418,8 @@ resolved owner, relevant scope, authority, canonical routing state, and source
 revision remain compatible. An unchanged source revision alone is insufficient.
 ### 5.6 Bounded multi-owner continuation control plane
 The optional continuation control plane is defined in
-`AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`.  It adds an
+`docs/standards/AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`.
+It adds an
 auditable route trace, acceptance progress, and independent continuation
 guards without creating a parallel state machine or changing the existing
 status namespaces.  Its guard thresholds are named parameters, not canonical
@@ -954,11 +955,12 @@ separate issue/PR) -> Phase 3 (artifact pilot) -> Phase 4 (Analytics pilot)
 Completion of this Phase 1 package does not authorize: pilot execution,
 semantic enforcement, CI blocking, merge, deploy, or production adoption.
 
-## From: `AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`
+## From: `docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`
 
 # Autonomous Execution Extension Contract
 Status: Phase 1 — normative package and declarative contract.
-Companion to: `AUTONOMOUS_EXECUTION_STANDARD.md` (canonical, `[AI OS]`-owned).
+Companion to: `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` (canonical,
+`[AI OS]`-owned).
 This is the one generic interface that a project-specific execution
 extension must implement. It is a template/contract, not a set of already
 adopted per-project extensions. Writing and adopting an actual extension for
@@ -1005,10 +1007,12 @@ closure_review_requirements:
   the state machine, defect model, or schema.
 ## 4. Precedence inside an extension
 An extension sits between the task package and the canonical standard in
-the precedence order defined in `AUTONOMOUS_EXECUTION_STANDARD.md` Section 1:
+the precedence order defined in
+`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` Section 1:
 ```text
 ... project instructions -> project-specific execution extension
--> AUTONOMOUS_EXECUTION_STANDARD.md -> supporting playbooks/templates/examples
+-> docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md
+-> supporting playbooks/templates/examples
 ```
 Where the extension and the canonical standard disagree on a limit, the
 stricter one applies.
@@ -1047,7 +1051,7 @@ the Phase 2 pilot task for the first project that actually adopts an
 extension, so it can follow that project's live Knowledge conventions
 instead of being guessed here.
 
-## From: `AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`
+## From: `docs/standards/AUTONOMOUS_EXECUTION_CONTINUATION_CONTROL_PLANE_CONTRACT.md`
 
 # AES Continuation Control Plane Contract
 Status: normative continuation semantics; implementation is additive and
@@ -1127,13 +1131,13 @@ Do not use this pattern to justify autonomous retrieval:
 ## Autonomous Execution Standard
 `[AI OS]` is the canonical owner of the Autonomous Execution Standard (AES).
 Execution across all projects now also follows the canonical loop defined in
-`AUTONOMOUS_EXECUTION_STANDARD.md` at the repo root: requirements -> execution
+`docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` at the repo root: requirements -> execution
 -> validation -> defect registration -> corrective action -> affected-scope
 rerun -> revalidation -> scope acceptance -> final evidence. It does not
 replace Goal Mode, routing, autonomy policy, or the merge policy in
 `GOAL_MODE.md`; it connects them into one closed loop, and the stricter rule
 wins on any conflict. `[AI OS]` also owns the generic project-extension
-interface in `AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`, which a project
+interface in `docs/standards/AUTONOMOUS_EXECUTION_EXTENSION_CONTRACT.md`, which a project
 implements to add domain-specific defect subtypes, evidence, and acceptance
 scopes without restating the canonical state machine or schema.
 New AES v2 records require Closure Review: it rechecks original goal, scope,

ChatGPT/[Analytics]/Knowledge/ANALYTICAL_REASONING_STANDARD.md

@@ -4,7 +4,7 @@
 
 This standard adds a bounded reasoning-control layer to the existing `[Analytics]` workflow. It does not replace or redefine the Data Contract, RAW, `stage_main_full`, `mart_main_full`, `mart_main_tz` / `mart_main_compact`, deterministic calculations, chart sourcing, memo pipeline, QA / Judge, acceptance, or Codex handoff boundary.
 
-Execution remains governed by the canonical repo-root `AUTONOMOUS_EXECUTION_STANDARD.md`. `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. This standard structures analytical reasoning and method selection only; it does not fork, weaken, or duplicate AES execution states, corrective-loop limits, stop conditions, rollback, acceptance, or external-action authority.
+Execution remains governed by the canonical repo-root `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`. `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. This standard structures analytical reasoning and method selection only; it does not fork, weaken, or duplicate AES execution states, corrective-loop limits, stop conditions, rollback, acceptance, or external-action authority.
 
 New AES execution records use the canonical v2 contract and its required
 Closure Review. Historical v1 records remain read-only evidence, not a path

ChatGPT/[Analytics]/Knowledge/ANALYTICS_03_TECHNIQUES_AND_CHARTS_BUNDLE_SEMANTICS.md

@@ -53,7 +53,7 @@ Every new non-CORE method uses the existing `trigger_type`, `trigger_rule`, `tri
 
 ## Boundary and flow
 This is a bounded reasoning-control extension; it does not replace Data Contract, RAW, STAGE, MART, deterministic calculations, chart sourcing, memo, QA/Judge, acceptance, or handoff. `mart_main_full` remains evidence/reuse; compact marts remain management-facing.
-Execution remains governed by repo-root `AUTONOMOUS_EXECUTION_STANDARD.md`; `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. Reasoning control structures method selection only; it does not fork AES states, correction limits, stop/rollback/acceptance, or external authority. It is not an autonomous agent or an independent retry/self-improvement loop. Closure Review uses existing Data/Analysis/Output Judge, reconciliation, unsupported-conclusion, and artifact-freshness controls; the Analytics extension remains authoritative for domain detail.
+Execution remains governed by repo-root `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`; `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. Reasoning control structures method selection only; it does not fork AES states, correction limits, stop/rollback/acceptance, or external authority. It is not an autonomous agent or an independent retry/self-improvement loop. Closure Review uses existing Data/Analysis/Output Judge, reconciliation, unsupported-conclusion, and artifact-freshness controls; the Analytics extension remains authoritative for domain detail.
 LLM reasoning != deterministic execution
 TASK PROFILE → INTENT → deterministic-first minimum methods → prerequisite gate
 → existing execution → preliminary evidence → explanation challenge when material

ChatGPT/[Analytics]/Knowledge/GOVERNANCE_AND_ANTI_PATTERNS.md

@@ -11,7 +11,7 @@
 - Method eligibility and triggers follow `ANALYTICAL_TECHNIQUES.md`; the LLM cannot silently override the registry.
 - Reasoning cannot substitute for deterministic execution or missing prerequisites.
 - Claim strength cannot exceed final evidence sufficiency.
-- `AUTONOMOUS_EXECUTION_STANDARD.md` remains canonical execution governance; `ANALYTICS_EXTENSION.md` supplies domain-specific constraints without creating a second execution framework.
+- `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` remains canonical execution governance; `ANALYTICS_EXTENSION.md` supplies domain-specific constraints without creating a second execution framework.
 
 ## Evidence labels
 

ChatGPT/[Analytics]/Knowledge_Bundles/ANALYTICS_03_TECHNIQUES_AND_CHARTS.md

@@ -21,7 +21,7 @@ ChatGPT Project Sources / Knowledge for `[Analytics]`.
 - production_promotion: no, unless explicitly accepted elsewhere
 - bundle_type: generated compact upload artifact
 - source_of_truth: declared granular source files
-- source_fingerprint: sha256:0774d47f87513d49e497622d1bbf6bcd40d1d270db84c94dddf284f5ba260d2c
+- source_fingerprint: sha256:7eb89642dffbb0ee2feaa8ee5a157d5d12052515abf06f2b07676618d0262612
 - generator: scripts/build_knowledge_bundles.py
 
 ---
@@ -194,7 +194,7 @@ Use only the minimum sufficient method set. A method must be able to materially
 # Analytical Reasoning Standard — P0
 ## Purpose and boundary
 This standard adds a bounded reasoning-control layer to the existing `[Analytics]` workflow. It does not replace or redefine the Data Contract, RAW, `stage_main_full`, `mart_main_full`, `mart_main_tz` / `mart_main_compact`, deterministic calculations, chart sourcing, memo pipeline, QA / Judge, acceptance, or Codex handoff boundary.
-Execution remains governed by the canonical repo-root `AUTONOMOUS_EXECUTION_STANDARD.md`. `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. This standard structures analytical reasoning and method selection only; it does not fork, weaken, or duplicate AES execution states, corrective-loop limits, stop conditions, rollback, acceptance, or external-action authority.
+Execution remains governed by the canonical repo-root `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`. `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. This standard structures analytical reasoning and method selection only; it does not fork, weaken, or duplicate AES execution states, corrective-loop limits, stop conditions, rollback, acceptance, or external-action authority.
 New AES execution records use the canonical v2 contract and its required
 Closure Review. Historical v1 records remain read-only evidence, not a path
 for new successful Analytics work.
@@ -869,7 +869,7 @@ Every new non-CORE method uses the existing `trigger_type`, `trigger_rule`, `tri
 ## Legacy section: `ChatGPT/[Analytics]/Knowledge/ANALYTICAL_REASONING_STANDARD.md`
 ## Boundary and flow
 This is a bounded reasoning-control extension; it does not replace Data Contract, RAW, STAGE, MART, deterministic calculations, chart sourcing, memo, QA/Judge, acceptance, or handoff. `mart_main_full` remains evidence/reuse; compact marts remain management-facing.
-Execution remains governed by repo-root `AUTONOMOUS_EXECUTION_STANDARD.md`; `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. Reasoning control structures method selection only; it does not fork AES states, correction limits, stop/rollback/acceptance, or external authority. It is not an autonomous agent or an independent retry/self-improvement loop. Closure Review uses existing Data/Analysis/Output Judge, reconciliation, unsupported-conclusion, and artifact-freshness controls; the Analytics extension remains authoritative for domain detail.
+Execution remains governed by repo-root `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md`; `docs/autonomous_execution/extensions/ANALYTICS_EXTENSION.md` remains authoritative for Analytics-specific AES requirements. Reasoning control structures method selection only; it does not fork AES states, correction limits, stop/rollback/acceptance, or external authority. It is not an autonomous agent or an independent retry/self-improvement loop. Closure Review uses existing Data/Analysis/Output Judge, reconciliation, unsupported-conclusion, and artifact-freshness controls; the Analytics extension remains authoritative for domain detail.
 LLM reasoning != deterministic execution
 TASK PROFILE → INTENT → deterministic-first minimum methods → prerequisite gate
 → existing execution → preliminary evidence → explanation challenge when material

ChatGPT/[Analytics]/Knowledge_Bundles/ANALYTICS_05_QA_GOVERNANCE_ROUTING.md

@@ -23,7 +23,7 @@ ChatGPT Project Sources / Knowledge for `[Analytics]`.
 - production_promotion: no, unless explicitly accepted elsewhere
 - bundle_type: generated compact upload artifact
 - source_of_truth: declared granular source files
-- source_fingerprint: sha256:39b7a92d62453c2e21df75e95603b0b7e2b47a644001c098914fde1c75790def
+- source_fingerprint: sha256:c6eed25af74b4722a385d0603c129c9aa0dcbe30cafc8efe24e01fa6561c8ddf
 - generator: scripts/build_knowledge_bundles.py
 
 ---
@@ -360,7 +360,7 @@ AI OS gives evidence and patterns. `[Analytics]` applies them only when they aff
 - Method eligibility and triggers follow `ANALYTICAL_TECHNIQUES.md`; the LLM cannot silently override the registry.
 - Reasoning cannot substitute for deterministic execution or missing prerequisites.
 - Claim strength cannot exceed final evidence sufficiency.
-- `AUTONOMOUS_EXECUTION_STANDARD.md` remains canonical execution governance; `ANALYTICS_EXTENSION.md` supplies domain-specific constraints without creating a second execution framework.
+- `docs/standards/AUTONOMOUS_EXECUTION_STANDARD.md` remains canonical execution governance; `ANALYTICS_EXTENSION.md` supplies domain-specific constraints without creating a second execution framework.
 ## Evidence labels
 Use:
 ```text