diff --git a/apps/api/migrations/adr090_asset_inventory_foundation.sql b/apps/api/migrations/adr090_asset_inventory_foundation.sql
new file mode 100644
index 00000000..00d01615
--- /dev/null
+++ b/apps/api/migrations/adr090_asset_inventory_foundation.sql
@@ -0,0 +1,607 @@
+-- ADR-090: 監控盲區治理 + 資產盤點 × 7 項自動化覆蓋矩陣永久化 DB
+-- 建立時間: 2026-04-18 下午 (台北時區)
+-- 建立者: ogt + Claude Opus 4.7 (1M context)（亞太）
+--
+-- 上游:
+--   - 主戰略: docs/superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md §5.2
+--   - ADR: docs/adr/ADR-090-monitoring-blindspot-governance.md
+--   - MEMORY: project_blindspot_governance.md
+--
+-- 設計說明:
+--   本檔建立 11 張表作為 AWOOOI L4 AIOps 的資產盤點 + 自動化覆蓋 + AI 協作稽核地基。
+--   目標: 把治理從 Markdown 搬進 PostgreSQL，讓 AI 四分工 (OpenClaw × NemoTron ×
+--         Hermes × Claude LLM) 在結構化資料上做決策，且每次動作必留 trail。
+--
+--   對應七大自動化引擎:
+--     E1 自動監控 / E2 自動告警 / E3 自動建規則 / E4 自動匹配
+--     E5 自動 Playbook / E6 自動修復 / E7 自動 KM
+--
+-- 執行順序:
+--   Step 0: pgcrypto extension (gen_random_uuid 需要)
+--   Step 1:  asset_inventory         — 全景資產主表
+--   Step 2:  asset_discovery_run     — 每次盤點 header
+--   Step 3:  asset_coverage_snapshot — 資產 × 7 自動化覆蓋矩陣
+--   Step 4:  asset_relationship      — 資產依賴圖 (爆炸半徑)
+--   Step 5:  alert_rule_catalog      — 告警規則本身即資產
+--   Step 6:  asset_change_event      — 資產變化追蹤
+--   Step 7:  asset_compliance_snapshot — SSL/CVE/secret/backup 合規
+--   Step 8:  host_capacity_snapshot  — 主機容量快照 (NemoTron 每日 02:00 寫)
+--   Step 9:  capacity_violation_event — 配額違規
+--   Step 10: automation_operation_log — 所有 AI 自動化動作稽核主表 🔴
+--   Step 11: ai_collaboration_trace  — 多 Agent 協作逐步 (辯證歷程)
+--   Step 12: 驗收查詢 (comment-only)
+--
+-- Idempotent 鐵律:
+--   - CREATE TABLE IF NOT EXISTS
+--   - CREATE INDEX IF NOT EXISTS
+--   - CHECK constraint 寫在 CREATE TABLE 內,依賴 IF NOT EXISTS 保護
+--   - 本檔可重複執行安全 (rerun 不會破壞既有資料)
+--
+-- 回滾:
+--   DROP TABLE IF EXISTS ai_collaboration_trace, automation_operation_log,
+--       capacity_violation_event, host_capacity_snapshot, asset_compliance_snapshot,
+--       asset_change_event, alert_rule_catalog, asset_relationship,
+--       asset_coverage_snapshot, asset_discovery_run, asset_inventory CASCADE;
+--
+-- ============================================================================
+-- Step 0: pgcrypto extension (gen_random_uuid)
+-- ============================================================================
+
+CREATE EXTENSION IF NOT EXISTS pgcrypto;
+
+
+-- ============================================================================
+-- Step 1: asset_inventory — 全景資產主表
+-- 用途: 主機 / 容器 / K8s workload / DB / 網站 / API / 套件 / 日誌 / KM / 前端 /
+--       後端 / 容器 / Gitea / CI-CD 全部無例外
+-- 主寫者: scanner (asset_discovery) + NemoTron (capacity 欄位)
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_inventory (
+    asset_id              BIGSERIAL       PRIMARY KEY,
+    asset_key             TEXT            NOT NULL UNIQUE,
+    asset_type            TEXT            NOT NULL,
+    parent_asset_id       BIGINT          REFERENCES asset_inventory(asset_id),
+    environment           TEXT            NOT NULL DEFAULT 'prod',
+    host                  TEXT,
+    namespace             TEXT,
+    name                  TEXT            NOT NULL,
+    metadata              JSONB           NOT NULL DEFAULT '{}'::jsonb,
+    tags                  TEXT[]          NOT NULL DEFAULT '{}',
+    owner_team            TEXT,
+    criticality           TEXT,
+    data_classification   TEXT,
+    external              BOOLEAN         NOT NULL DEFAULT false,
+    lifecycle_state       TEXT            NOT NULL DEFAULT 'active',
+    source_repo           TEXT,
+    source_commit_sha     TEXT,
+
+    -- 容量欄位 (Layer 4 AI 巡檢用)
+    cpu_avg_7d            NUMERIC(5,2),
+    mem_avg_7d            NUMERIC(5,2),
+    capacity_headroom     NUMERIC(5,2),
+    resource_limits       JSONB,
+    resource_requests     JSONB,
+    quota_violation_count INT             NOT NULL DEFAULT 0,
+    sla_target            JSONB,
+    cost_monthly_usd      NUMERIC(10,2),
+
+    -- 生命週期時間戳
+    first_seen_at         TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    last_seen_at          TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    decommissioned_at     TIMESTAMPTZ,
+    created_at            TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    updated_at            TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT asset_inventory_criticality_valid
+        CHECK (criticality IS NULL OR criticality IN ('P0','P1','P2','P3')),
+    CONSTRAINT asset_inventory_data_class_valid
+        CHECK (data_classification IS NULL OR data_classification IN
+            ('public','internal','sensitive','secret')),
+    CONSTRAINT asset_inventory_lifecycle_valid
+        CHECK (lifecycle_state IN
+            ('planned','provisioning','active','degraded','deprecated','decommissioned')),
+    CONSTRAINT asset_inventory_type_valid
+        CHECK (asset_type IN (
+            'host','container','k8s_workload','k8s_resource','database','table',
+            'website','api_endpoint','package','log_stream','km_entry',
+            'frontend','backend','ci_pipeline','gitea_repo','monitoring_target',
+            'secret','volume','network','certificate','scheduled_job',
+            'message_queue','cache','dashboard','ai_agent','llm_model',
+            'third_party_service','backup_target'
+        ))
+);
+
+COMMENT ON TABLE asset_inventory IS
+    'ADR-090: 全景資產主表。每一個主機/容器/K8s workload/DB/網站/API/套件/...都有一筆,跨 run 沿用同 asset_id。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_inventory_type_host
+    ON asset_inventory(asset_type, host);
+CREATE INDEX IF NOT EXISTS idx_asset_inventory_env_lifecycle
+    ON asset_inventory(environment, lifecycle_state);
+CREATE INDEX IF NOT EXISTS idx_asset_inventory_metadata_gin
+    ON asset_inventory USING GIN (metadata);
+CREATE INDEX IF NOT EXISTS idx_asset_inventory_tags_gin
+    ON asset_inventory USING GIN (tags);
+CREATE INDEX IF NOT EXISTS idx_asset_inventory_active_last_seen
+    ON asset_inventory(last_seen_at DESC)
+    WHERE lifecycle_state = 'active';
+-- 註: partial index 只索引 active 資產,按最近出現時間排序
+
+
+-- ============================================================================
+-- Step 2: asset_discovery_run — 每次盤點 header
+-- 用途: 記錄每次全景掃描的起止時間、掃描範圍、掃到什麼、新增/消失多少
+-- 觸發: cron (每日) / ai (proactive_inspector) / human (手動) / incident
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_discovery_run (
+    run_id              UUID            PRIMARY KEY DEFAULT gen_random_uuid(),
+    triggered_by        TEXT            NOT NULL,
+    scope               TEXT[]          NOT NULL,
+    scan_depth          TEXT            NOT NULL DEFAULT 'shallow',
+    host_filter         TEXT[],
+    started_at          TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    ended_at            TIMESTAMPTZ,
+    status              TEXT            NOT NULL,
+    total_assets        INT,
+    new_assets          INT             NOT NULL DEFAULT 0,
+    modified_assets     INT             NOT NULL DEFAULT 0,
+    disappeared_assets  INT             NOT NULL DEFAULT 0,
+    tools_used          JSONB,
+    duration_ms         INT,
+    error               TEXT,
+    summary             JSONB,
+
+    CONSTRAINT asset_discovery_run_status_valid
+        CHECK (status IN ('running','success','partial','failed','aborted')),
+    CONSTRAINT asset_discovery_run_scan_depth_valid
+        CHECK (scan_depth IN ('shallow','deep','full'))
+);
+
+COMMENT ON TABLE asset_discovery_run IS
+    'ADR-090: 每次資產盤點的 header。run_id 作為下游 snapshot/event/change 的關聯主鍵。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_discovery_run_started
+    ON asset_discovery_run(started_at DESC);
+CREATE INDEX IF NOT EXISTS idx_asset_discovery_run_status
+    ON asset_discovery_run(status) WHERE status IN ('running','failed','partial');
+
+
+-- ============================================================================
+-- Step 3: asset_coverage_snapshot — 資產 × 7 項自動化 覆蓋矩陣
+-- 用途: 每個資產在 7 個自動化維度上的覆蓋狀態 (green/yellow/red)
+-- 鐵律: 每次 discovery_run 為每個 asset 寫 7 筆 (7 dimensions)
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_coverage_snapshot (
+    snapshot_id         BIGSERIAL       PRIMARY KEY,
+    run_id              UUID            NOT NULL REFERENCES asset_discovery_run(run_id) ON DELETE CASCADE,
+    asset_id            BIGINT          NOT NULL REFERENCES asset_inventory(asset_id),
+    dimension           TEXT            NOT NULL,
+    coverage_status     TEXT            NOT NULL,
+    evidence            JSONB           NOT NULL DEFAULT '{}'::jsonb,
+    gap_reason          TEXT,
+    recommended_action  TEXT,
+    confidence          NUMERIC(3,2),
+    detected_by         TEXT            NOT NULL,
+    created_at          TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT asset_coverage_snapshot_dimension_valid
+        CHECK (dimension IN (
+            'auto_monitoring','auto_alerting','auto_rule_creation',
+            'auto_rule_matching','auto_playbook','auto_remediation','auto_km_creation'
+        )),
+    CONSTRAINT asset_coverage_snapshot_status_valid
+        CHECK (coverage_status IN ('green','yellow','red','unknown')),
+    CONSTRAINT asset_coverage_snapshot_unique
+        UNIQUE (run_id, asset_id, dimension)
+);
+
+COMMENT ON TABLE asset_coverage_snapshot IS
+    'ADR-090: 計分卡。查 red COUNT 即覆蓋率 SLO。evidence 欄位串 playbook_id/km_entry_id/rule_name。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_coverage_snapshot_asset_dim
+    ON asset_coverage_snapshot(asset_id, dimension);
+CREATE INDEX IF NOT EXISTS idx_asset_coverage_snapshot_red_yellow
+    ON asset_coverage_snapshot(coverage_status)
+    WHERE coverage_status IN ('red','yellow');
+CREATE INDEX IF NOT EXISTS idx_asset_coverage_snapshot_run
+    ON asset_coverage_snapshot(run_id);
+
+
+-- ============================================================================
+-- Step 4: asset_relationship — 資產依賴圖 (爆炸半徑必需)
+-- 用途: 記錄資產之間的 depends_on / calls / stores_data_in / backs_up_to 關係
+-- AI 用途: OpenClaw 計算 blast_radius 時查這張表
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_relationship (
+    relationship_id     BIGSERIAL       PRIMARY KEY,
+    from_asset_id       BIGINT          NOT NULL REFERENCES asset_inventory(asset_id),
+    to_asset_id         BIGINT          NOT NULL REFERENCES asset_inventory(asset_id),
+    relationship_type   TEXT            NOT NULL,
+    strength            NUMERIC(3,2),
+    metadata            JSONB,
+    first_detected_at   TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    last_verified_at    TIMESTAMPTZ,
+    is_active           BOOLEAN         NOT NULL DEFAULT true,
+
+    CONSTRAINT asset_relationship_type_valid
+        CHECK (relationship_type IN (
+            'depends_on','calls','stores_data_in','backs_up_to',
+            'routes_to','authenticates_via','monitors','alerts_to','logs_to'
+        )),
+    CONSTRAINT asset_relationship_strength_valid
+        CHECK (strength IS NULL OR (strength >= 0 AND strength <= 1)),
+    CONSTRAINT asset_relationship_unique
+        UNIQUE (from_asset_id, to_asset_id, relationship_type),
+    CONSTRAINT asset_relationship_no_self_loop
+        CHECK (from_asset_id <> to_asset_id)
+);
+
+COMMENT ON TABLE asset_relationship IS
+    'ADR-090: 資產依賴圖。AI 計算爆炸半徑必讀。edge 而非 tree,支援多重關係。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_relationship_from
+    ON asset_relationship(from_asset_id) WHERE is_active;
+CREATE INDEX IF NOT EXISTS idx_asset_relationship_to
+    ON asset_relationship(to_asset_id) WHERE is_active;
+CREATE INDEX IF NOT EXISTS idx_asset_relationship_type
+    ON asset_relationship(relationship_type);
+
+
+-- ============================================================================
+-- Step 5: alert_rule_catalog — 告警規則本身即資產
+-- 用途: 把 alert_rules.yaml 升級為 DB-driven;記錄誰創的 / 何時 / 效能 / 生死
+-- AI 用途: Hermes 做 noise_rate 分析 / 提建議 retire 低品質規則
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS alert_rule_catalog (
+    rule_id                 BIGSERIAL       PRIMARY KEY,
+    rule_name               TEXT            NOT NULL UNIQUE,
+    source                  TEXT            NOT NULL,
+    expr                    TEXT            NOT NULL,
+    duration_seconds        INT,
+    severity                TEXT,
+    labels                  JSONB,
+    annotations             JSONB,
+    linked_asset_ids        BIGINT[],
+    created_by_agent        TEXT,
+
+    -- 規則品質追蹤
+    true_positive_count     INT             NOT NULL DEFAULT 0,
+    false_positive_count    INT             NOT NULL DEFAULT 0,
+    noise_rate              NUMERIC(5,2),
+    last_fired_at           TIMESTAMPTZ,
+
+    -- 信心與演化
+    confidence              NUMERIC(3,2),
+    review_status           TEXT,
+    superseded_by_rule_id   BIGINT          REFERENCES alert_rule_catalog(rule_id),
+
+    created_at              TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    updated_at              TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT alert_rule_catalog_source_valid
+        CHECK (source IN ('yaml_hardcoded','ai_generated','human_written','playbook_derived')),
+    CONSTRAINT alert_rule_catalog_review_valid
+        CHECK (review_status IS NULL OR review_status IN
+            ('draft','approved','deprecated','retired'))
+);
+
+COMMENT ON TABLE alert_rule_catalog IS
+    'ADR-090: 告警規則即一等資產。支援規則演化 (ai_generated) 與替代鏈 (superseded_by)。';
+
+CREATE INDEX IF NOT EXISTS idx_alert_rule_catalog_source
+    ON alert_rule_catalog(source);
+CREATE INDEX IF NOT EXISTS idx_alert_rule_catalog_assets_gin
+    ON alert_rule_catalog USING GIN (linked_asset_ids);
+CREATE INDEX IF NOT EXISTS idx_alert_rule_catalog_review
+    ON alert_rule_catalog(review_status) WHERE review_status IS NOT NULL;
+
+
+-- ============================================================================
+-- Step 6: asset_change_event — 資產變化追蹤 (diff between runs)
+-- 用途: 兩次 discovery_run 之間的 delta。新增/消失/修改/覆蓋率變化
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_change_event (
+    event_id        BIGSERIAL       PRIMARY KEY,
+    run_id          UUID            NOT NULL REFERENCES asset_discovery_run(run_id),
+    asset_id        BIGINT          REFERENCES asset_inventory(asset_id),
+    change_type     TEXT            NOT NULL,
+    before_state    JSONB,
+    after_state     JSONB,
+    diff            JSONB,
+    detected_at     TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    ai_analysis     TEXT,
+
+    CONSTRAINT asset_change_event_type_valid
+        CHECK (change_type IN (
+            'asset_added','asset_removed','asset_modified',
+            'coverage_improved','coverage_degraded',
+            'criticality_changed','owner_changed','lifecycle_changed'
+        ))
+);
+
+COMMENT ON TABLE asset_change_event IS
+    'ADR-090: 資產變化追蹤。兩次掃描的 diff 明確落地,LLM 可加 ai_analysis 解讀。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_change_event_run
+    ON asset_change_event(run_id);
+CREATE INDEX IF NOT EXISTS idx_asset_change_event_asset_time
+    ON asset_change_event(asset_id, detected_at DESC);
+
+
+-- ============================================================================
+-- Step 7: asset_compliance_snapshot — 合規狀態 (SSL/CVE/secret/backup)
+-- 用途: 與 coverage 不同軸的合規追蹤。SSL cert 到期 / CVE 掃描 / secret 輪替
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS asset_compliance_snapshot (
+    snapshot_id             BIGSERIAL       PRIMARY KEY,
+    run_id                  UUID            REFERENCES asset_discovery_run(run_id),
+    asset_id                BIGINT          NOT NULL REFERENCES asset_inventory(asset_id),
+    dimension               TEXT            NOT NULL,
+    status                  TEXT            NOT NULL,
+    expires_at              TIMESTAMPTZ,
+    detail                  JSONB,
+    remediation_deadline    TIMESTAMPTZ,
+    detected_at             TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT asset_compliance_snapshot_dimension_valid
+        CHECK (dimension IN (
+            'ssl_cert_valid','cve_scan','secret_rotated','backup_tested',
+            'audit_log_enabled','access_reviewed','encryption_at_rest'
+        )),
+    CONSTRAINT asset_compliance_snapshot_status_valid
+        CHECK (status IN ('compliant','warning','violation','unknown'))
+);
+
+COMMENT ON TABLE asset_compliance_snapshot IS
+    'ADR-090: 合規狀態快照。與 coverage 不同軸,SSL/CVE/secret/backup 專用。';
+
+CREATE INDEX IF NOT EXISTS idx_asset_compliance_snapshot_asset_dim
+    ON asset_compliance_snapshot(asset_id, dimension);
+CREATE INDEX IF NOT EXISTS idx_asset_compliance_snapshot_expiring
+    ON asset_compliance_snapshot(expires_at)
+    WHERE expires_at IS NOT NULL;
+CREATE INDEX IF NOT EXISTS idx_asset_compliance_snapshot_violations
+    ON asset_compliance_snapshot(status)
+    WHERE status IN ('warning','violation');
+
+
+-- ============================================================================
+-- Step 8: host_capacity_snapshot — 主機容量快照
+-- 用途: NemoTron 每日 02:00 台北 自主容量巡檢寫入
+-- Layer 4 核心表。hermes 做預測,openclaw 產建議,全寫這張
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS host_capacity_snapshot (
+    snapshot_id             BIGSERIAL       PRIMARY KEY,
+    host                    TEXT            NOT NULL,
+    captured_at             TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    load1                   NUMERIC(6,2),
+    load5                   NUMERIC(6,2),
+    load15                  NUMERIC(6,2),
+    cpu_used_pct            NUMERIC(5,2),
+    cpu_iowait_pct          NUMERIC(5,2),
+    mem_used_pct            NUMERIC(5,2),
+    swap_used_pct           NUMERIC(5,2),
+    disk_used_pct           JSONB,
+    container_count         INT,
+    k8s_pod_count           INT,
+    top_cpu_offenders       JSONB,
+    top_mem_offenders       JSONB,
+    headroom_pct            NUMERIC(5,2),
+    ai_verdict              TEXT,
+    ai_reasoning            TEXT,
+    recommended_actions     JSONB,
+    written_by_agent        TEXT            NOT NULL,
+
+    CONSTRAINT host_capacity_snapshot_verdict_valid
+        CHECK (ai_verdict IS NULL OR ai_verdict IN ('safe','warning','critical','unknown'))
+);
+
+COMMENT ON TABLE host_capacity_snapshot IS
+    'ADR-090: NemoTron 每日主機容量巡檢結果。Layer 4 AI 自主治理核心表。';
+
+CREATE INDEX IF NOT EXISTS idx_host_capacity_snapshot_host_time
+    ON host_capacity_snapshot(host, captured_at DESC);
+CREATE INDEX IF NOT EXISTS idx_host_capacity_snapshot_critical
+    ON host_capacity_snapshot(ai_verdict)
+    WHERE ai_verdict IN ('warning','critical');
+
+
+-- ============================================================================
+-- Step 9: capacity_violation_event — 配額違規事件
+-- 用途: 記錄任何「缺 limit」「超 request」「主機飽和」的違規
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS capacity_violation_event (
+    event_id            BIGSERIAL       PRIMARY KEY,
+    asset_id            BIGINT          REFERENCES asset_inventory(asset_id),
+    host                TEXT,
+    violation_type      TEXT            NOT NULL,
+    threshold           NUMERIC(10,2),
+    actual_value        NUMERIC(10,2),
+    detected_at         TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+    auto_action         TEXT,
+    auto_action_op_id   UUID,
+    human_override      TEXT,
+    resolved_at         TIMESTAMPTZ,
+
+    CONSTRAINT capacity_violation_event_type_valid
+        CHECK (violation_type IN (
+            'no_limit_set','over_request','over_limit','host_saturation',
+            'over_sla_budget','unauthorized_new_deploy'
+        ))
+);
+
+COMMENT ON TABLE capacity_violation_event IS
+    'ADR-090: 配額違規稽核。每次 AI 偵測到資產無 limit/主機飽和/未授權部署 都寫一筆。';
+
+CREATE INDEX IF NOT EXISTS idx_capacity_violation_event_asset_time
+    ON capacity_violation_event(asset_id, detected_at DESC);
+CREATE INDEX IF NOT EXISTS idx_capacity_violation_event_unresolved
+    ON capacity_violation_event(detected_at DESC)
+    WHERE resolved_at IS NULL;
+
+
+-- ============================================================================
+-- Step 10: automation_operation_log — 所有 AI 自動化動作稽核主表 🔴
+-- 鐵律: 每一個 AI 自動化動作都必須寫一筆。缺筆 = 治理失效
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS automation_operation_log (
+    op_id               UUID            PRIMARY KEY DEFAULT gen_random_uuid(),
+    operation_type      TEXT            NOT NULL,
+    asset_id            BIGINT          REFERENCES asset_inventory(asset_id),
+    incident_id         BIGINT,
+    run_id              UUID            REFERENCES asset_discovery_run(run_id),
+    actor               TEXT            NOT NULL,
+    input               JSONB           NOT NULL DEFAULT '{}'::jsonb,
+    output              JSONB           NOT NULL DEFAULT '{}'::jsonb,
+    dry_run_result      JSONB,
+    status              TEXT            NOT NULL,
+    error               TEXT,
+    duration_ms         INT,
+    tokens_in           INT,
+    tokens_out          INT,
+    cost_usd            NUMERIC(10,6),
+    budget_bucket       TEXT,
+    parent_op_id        UUID            REFERENCES automation_operation_log(op_id),
+    retry_count         INT             NOT NULL DEFAULT 0,
+    retry_of_op_id      UUID            REFERENCES automation_operation_log(op_id),
+    stderr_feed_back    TEXT,
+    tags                TEXT[],
+    created_at          TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT automation_operation_log_type_valid
+        CHECK (operation_type IN (
+            'monitor_configured','monitor_removed',
+            'alert_fired','alert_suppressed','alert_routed',
+            'rule_created','rule_updated','rule_matched','rule_rejected','rule_deprecated',
+            'playbook_generated','playbook_updated','playbook_executed',
+            'remediation_executed','remediation_verified','remediation_rolled_back',
+            'self_correction_attempted',
+            'km_created','km_updated','km_linked',
+            'asset_discovered','coverage_recalculated',
+            'capacity_recommendation','quota_enforced'
+        )),
+    CONSTRAINT automation_operation_log_status_valid
+        CHECK (status IN ('pending','success','failed','dry_run','rolled_back'))
+);
+
+COMMENT ON TABLE automation_operation_log IS
+    'ADR-090: 所有 AI 自動化動作稽核主表。retry_of_op_id + stderr_feed_back 支援引擎 4 閉環。';
+
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_type_time
+    ON automation_operation_log(operation_type, created_at DESC);
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_asset_time
+    ON automation_operation_log(asset_id, created_at DESC);
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_incident
+    ON automation_operation_log(incident_id)
+    WHERE incident_id IS NOT NULL;
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_actor_time
+    ON automation_operation_log(actor, created_at DESC);
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_retry
+    ON automation_operation_log(retry_of_op_id)
+    WHERE retry_of_op_id IS NOT NULL;
+CREATE INDEX IF NOT EXISTS idx_automation_operation_log_tags_gin
+    ON automation_operation_log USING GIN (tags);
+
+
+-- ============================================================================
+-- Step 11: ai_collaboration_trace — 多 Agent 協作逐步 (LLM × OpenClaw × NemoTron × Hermes)
+-- 用途: 每個 automation_operation_log 背後的 N 步 AI 決策過程
+-- 最寶貴的語料: challenged_by + accepted 支援 RLHF fine-tune
+-- ============================================================================
+
+CREATE TABLE IF NOT EXISTS ai_collaboration_trace (
+    trace_id                BIGSERIAL       PRIMARY KEY,
+    op_id                   UUID            NOT NULL REFERENCES automation_operation_log(op_id) ON DELETE CASCADE,
+    step_order              INT             NOT NULL,
+    agent                   TEXT            NOT NULL,
+    model                   TEXT,
+    system_prompt_version   TEXT,
+    prompt                  TEXT,
+    response                JSONB,
+    confidence              NUMERIC(3,2),
+    challenged_by           TEXT[],
+    accepted                BOOLEAN,
+    tokens_in               INT,
+    tokens_out              INT,
+    duration_ms             INT,
+    created_at              TIMESTAMPTZ     NOT NULL DEFAULT NOW(),
+
+    CONSTRAINT ai_collaboration_trace_unique_step
+        UNIQUE (op_id, step_order)
+);
+
+COMMENT ON TABLE ai_collaboration_trace IS
+    'ADR-090: AI 多 Agent 協作逐步紀錄。challenged_by + accepted = RLHF 訓練語料金礦。';
+
+CREATE INDEX IF NOT EXISTS idx_ai_collaboration_trace_op
+    ON ai_collaboration_trace(op_id, step_order);
+CREATE INDEX IF NOT EXISTS idx_ai_collaboration_trace_agent_time
+    ON ai_collaboration_trace(agent, created_at DESC);
+
+
+-- ============================================================================
+-- Step 12: 驗收查詢 (執行後手動跑,驗證 11 張表都到位)
+-- ============================================================================
+
+-- SELECT table_name
+-- FROM information_schema.tables
+-- WHERE table_schema = 'public'
+--   AND table_name IN (
+--       'asset_inventory',
+--       'asset_discovery_run',
+--       'asset_coverage_snapshot',
+--       'asset_relationship',
+--       'alert_rule_catalog',
+--       'asset_change_event',
+--       'asset_compliance_snapshot',
+--       'host_capacity_snapshot',
+--       'capacity_violation_event',
+--       'automation_operation_log',
+--       'ai_collaboration_trace'
+--   )
+-- ORDER BY table_name;
+-- -- 預期: 11 筆
+
+-- SELECT table_name, COUNT(*) AS column_count
+-- FROM information_schema.columns
+-- WHERE table_schema = 'public'
+--   AND table_name LIKE 'asset_%' OR table_name IN
+--       ('alert_rule_catalog','host_capacity_snapshot','capacity_violation_event',
+--        'automation_operation_log','ai_collaboration_trace')
+-- GROUP BY table_name
+-- ORDER BY table_name;
+
+-- SELECT conname, conrelid::regclass AS table_name
+-- FROM pg_constraint
+-- WHERE conrelid IN (
+--     'asset_inventory'::regclass,
+--     'asset_discovery_run'::regclass,
+--     'asset_coverage_snapshot'::regclass,
+--     'asset_relationship'::regclass,
+--     'alert_rule_catalog'::regclass,
+--     'asset_change_event'::regclass,
+--     'asset_compliance_snapshot'::regclass,
+--     'host_capacity_snapshot'::regclass,
+--     'capacity_violation_event'::regclass,
+--     'automation_operation_log'::regclass,
+--     'ai_collaboration_trace'::regclass
+-- ) AND contype = 'c'  -- CHECK constraints only
+-- ORDER BY table_name, conname;
+
+-- ============================================================================
+-- END OF MIGRATION adr090_asset_inventory_foundation.sql
+-- 預計新增物件: 11 tables + 33 indexes + 20 CHECK constraints + 3 UNIQUE + 16 FK references
+-- 依賴: pgcrypto extension (for gen_random_uuid)
+-- 影響資料: 無 (純 DDL, 不動現有表)
+-- 回滾: 見檔案頭部
+-- ============================================================================
diff --git a/docs/adr/ADR-090-monitoring-blindspot-governance.md b/docs/adr/ADR-090-monitoring-blindspot-governance.md
new file mode 100644
index 00000000..a7b08091
--- /dev/null
+++ b/docs/adr/ADR-090-monitoring-blindspot-governance.md
@@ -0,0 +1,104 @@
+# ADR-090: 監控盲區治理 + 資產盤點 × 7 項自動化覆蓋矩陣永久化 DB
+
+**日期**: 2026-04-18
+**狀態**: 提案（Schema 定稿，Migration 待執行）
+**上游**: [ADR-080 AI 自主化飛輪總覽](./ADR-080-ai-autonomy-flywheel-overview.md)、[2026-04-15 MASTER 藍圖](../superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md)
+**完整實施計畫**: [2026-04-18 盲區治理 + 容量自主化戰役](../superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md)
+
+## Context（背景）
+
+2026-04-18 MoWoooWorkDown 假警報 RCA 暴露三重結構性失守：
+
+1. **監控覆蓋不到三成** — Prometheus 僅 35 active targets / 58 alert rules，對照實際資產（主機、K3s workloads、40+ Sentry 容器、18 個 188 容器、Kafka、ClickHouse、Ollama GPU、120/121 K3s Masters、多個網站）嚴重不足
+2. **告警規則量錯維度** — `HostHighCpuLoad` 用 CPU idle% 測，結果 110 load=18.35 / 188 load=16.32 持續 13 天無觸發；cadvisor 狂吃 288% CPU 13 天無告警
+3. **MD/ADR/Memory 無強制力** — 過去寫了一堆治理文件，執行上仍持續偏移，因為沒有資料庫級的強制檢查
+
+MASTER Phase 0-6 的飛輪代碼全綠，但**在生產上沒真轉**：Playbook trust 實際動過幾次？MCP 24h 真 >500 嗎？RESTART 佔比真從 68% 降到 <40% 嗎？全部沒量過。
+
+## Decision（決策）
+
+建立 **11 張表的資產盤點 + 自動化覆蓋 + AI 協作稽核資料庫地基**，把治理從 Markdown 搬進 PostgreSQL，讓 AI 四分工（OpenClaw × NemoTron × Hermes × Claude LLM）在結構化資料上做決策，且每次動作必留 trail。
+
+### 11 張表總覽
+
+| # | 表 | 用途 | 主寫者 |
+|---|---|------|-------|
+| 1 | `asset_inventory` | 全景資產主表（主機/容器/K8s/DB/網站/API/套件…） | scanner + NemoTron |
+| 2 | `asset_discovery_run` | 每次盤點 header | cron / ai / human |
+| 3 | `asset_coverage_snapshot` | 資產 × 7 自動化 覆蓋矩陣 | scanner + OpenClaw |
+| 4 | `asset_relationship` | 資產依賴圖（爆炸半徑） | OpenClaw |
+| 5 | `alert_rule_catalog` | 告警規則本身即資產 | human + Hermes |
+| 6 | `asset_change_event` | 資產變化追蹤 | scanner diff |
+| 7 | `asset_compliance_snapshot` | SSL / CVE / secret / backup 合規 | scanner |
+| 8 | `host_capacity_snapshot` | 主機容量快照（每日 02:00） | **NemoTron** 🔴 |
+| 9 | `capacity_violation_event` | 配額違規 | 自動 |
+| 10 | `automation_operation_log` | **所有 AI 自動化動作稽核主表** | 全 Agent |
+| 11 | `ai_collaboration_trace` | 多 Agent 協作逐步（辯證歷程） | 全 Agent |
+
+### 七大自動化引擎對映
+
+每個資產在 `asset_coverage_snapshot.dimension` 有七格：
+- `auto_monitoring` — Prom/Log/Metric 有抓嗎
+- `auto_alerting` — 有對應 alert rule 嗎
+- `auto_rule_creation` — 規則是人寫還是 AI 產
+- `auto_rule_matching` — 告警來時 rule 有自動匹配嗎
+- `auto_playbook` — 有對應 Playbook 嗎
+- `auto_remediation` — 有自動修復能力嗎
+- `auto_km_creation` — 故障後自動進 KM 嗎
+
+### AI 四分工責任矩陣
+
+| 引擎 | 主責 | 輔責 |
+|-----|-----|------|
+| E1 自動監控 | **NemoTron** 每日容量巡檢 | Hermes 趨勢 |
+| E2 自動告警 | **OpenClaw** 分類分診 | Critic 挑戰 |
+| E3 自動建規則 | **Hermes** 分析師 | NemoTron 提案 |
+| E4 自動匹配 | **OpenClaw** embedding 檢索 | NemoTron 排序 |
+| E5 自動 Playbook | **NemoTron + Solver** | Reviewer |
+| E6 自動修復 | **Coordinator** + Critic + Verifier | stderr 回灌 |
+| E7 自動 KM | **Hermes** 敘事 | OpenClaw 摘要 |
+
+## Alternatives Considered（考慮過的替代）
+
+| 方案 | 為何否決 |
+|-----|---------|
+| A. 只改 alert_rules.yaml 補規則 | 治標不治本，無法追蹤「為何加/何時 retire」 |
+| B. 用 Grafana / Datadog 外部 SaaS | 不符合自託管偏好 + 歷程不進自家 DB |
+| C. 6 張表版（未加 relationship/alert_rule_catalog/compliance/capacity） | 爆炸半徑算不出、規則無演化記錄、配額違規無稽核 |
+| D. 用既有 incidents / knowledge_entries 湊合 | 這兩表是事件層，不是資產層，硬塞會搞垮 schema |
+
+## Consequences（後果）
+
+### Positive
+
+- **治理從 MD 升級為 DB 強制**：Gitea CI 可 query DB 做 PR reject
+- **AI 可跨資產做決策**：NemoTron 看 `host_capacity_snapshot` 自主提搬遷建議
+- **所有 AI 動作有稽核**：`automation_operation_log` + `ai_collaboration_trace` 形成完整 fine-tune 訓料
+- **盲區可量化**：`asset_coverage_snapshot.coverage_status='red'` COUNT 即覆蓋率 SLO
+- **規則可演化**：`alert_rule_catalog.noise_rate` + `superseded_by_rule_id` 支援 AI 漸進取代硬編規則
+
+### Negative
+
+- **11 張表 + 多 FK 增加 Migration 複雜度**（但分批回滾可行）
+- **PostgreSQL 188 寫入量提升**（現有低寫入，加這些每日寫 <10k 行，可忽略）
+- **Scanner 須寫新腳本**（`apps/api/scripts/audit_assets_v1.py`）
+- **現有 25 條 alert_rules.yaml 須 seed 進 alert_rule_catalog**（一次性，~1 小時）
+
+### Rollback
+
+- 每張表獨立 migration，可 `alembic downgrade -1` 一張張退
+- Feature flag `AIOPS_ASSET_INVENTORY_ENABLED` 預設 false，分階段開啟
+
+## Implementation（施工）
+
+詳見 [2026-04-18 盲區治理 + 容量自主化戰役](../superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md) §5 完整 DDL、§7 七階段實施步驟、§10 驗收指標。
+
+## Related
+
+- ADR-080 AI 自主化飛輪總覽
+- ADR-082 多 Agent 協作
+- ADR-083 學習閉環
+- ADR-084 異常偵測（trend_predictor）
+- ADR-086 修復抽象化（blast_radius_calculator）
+- MASTER §3.1 D1 感官縱深（MCP Tool Registry）
+- MASTER §4 7 層 × 6 維矩陣
diff --git a/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md b/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
index 0117952b..bcb84de5 100644
--- a/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
+++ b/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
@@ -1597,3 +1597,47 @@ Phase 6 完成後
 - [ ] Fine-tune JSONL ≥ 10 條（EvidenceSnapshot 累積 7 天後）
 
 **下一步：** 推 Gitea → CD 部署 → 呼叫 `POST /api/v1/learning/evolver/run` 完成合併演練 → 7 天生產監控
+
+---
+
+### 2026-04-18 下午 (台北) — Phase 7 啟動：盲區治理 + 容量自主化 — 新子藍圖分枝
+
+**觸發**：MoWoooWorkDown 假警報 RCA 深潛 → 暴露三重結構性失守（見下）
+
+**鐵證**：
+- 110 主機 load=18.35 / 188 主機 load=16.32 / cadvisor 狂吃 288% CPU **持續 13 天無告警**
+- Prometheus 僅 35 active targets / 58 alert rules — **覆蓋不到三成**
+- `HostHighCpuLoad` 規則用 `CPU idle < 80%` 測 — **量錯維度**（應測 load_avg / iowait）
+- K3s Master 120/121 + VIP 125 完全沒 node-exporter
+- Kafka / Ollama GPU / ClickHouse / Sentry 40+ 容器全黑箱
+
+**統帥戰略指令**（不可動搖）：
+1. 全景資產（主機/環境/服務/監控/工作/套件/DB/日誌/KM/前端/後端/API/容器/Gitea/CI-CD 無例外）× 七大自動化 × 永久化 DB
+2. AI 四分工：OpenClaw × NemoTron × Hermes × Claude LLM 辯證協作
+3. **所有自動化操作歷程必進 DB**，不靠 MD（MD 會漂移）
+
+**子藍圖誕生**：[`2026-04-18-blindspot-governance-capacity-l4.md`](./2026-04-18-blindspot-governance-capacity-l4.md)
+- 本檔（MASTER v2）維持 Phase 0-6 地位不變
+- 子藍圖為 **Phase 7**：盲區矯正 + 容量自主化
+- 子藍圖 §5 定義 11 張表 Schema（asset_inventory / asset_discovery_run / asset_coverage_snapshot / asset_relationship / alert_rule_catalog / asset_change_event / asset_compliance_snapshot / host_capacity_snapshot / capacity_violation_event / automation_operation_log / ai_collaboration_trace）
+- 子藍圖 §6 定義 4 層防禦（告警精準化 / 配額強制 / 主機搬遷 / AI 容量巡檢）
+- 子藍圖 §7 定義 7 Phase 實施順序（0a DNS 止血 → 0b cadvisor 降壓 → 1 Migration → 2 全景審計 → 3 規則升級 → 4 NemoTron 容量巡檢 → 5 配額強制 → 6 主機搬遷 → 7 引擎 3+4 閉環）
+
+**新增 ADR**：[`ADR-090`](../../adr/ADR-090-monitoring-blindspot-governance.md)
+
+**新增 HARD_RULES**：
+1. 監控工具必須被監控（cadvisor 案例）
+2. 配額即義務（缺 limit → CI reject）
+3. 主機容量警戒線（headroom < 0.2 三日 → 凍結新部署）
+4. 告警維度鐵律（必含 load_avg / iowait）
+5. IP 靜態綁定必登記技術債
+6. 監控覆蓋率 SLO（red 比例 < 20%）
+7. 新服務上線 Gate（先進 asset_inventory）
+
+**新增 Memory**：
+- `project_blindspot_governance.md` — 子藍圖跨 session 指針
+- `feedback_monitor_self_monitoring.md` — 🔴🔴🔴 監控工具必須被監控鐵律
+
+**當前 Blocker**：Phase 0a/0b 止血指令待統帥最終授權。110/188 load 持續超載，執行重審計前必須先降壓。
+
+**下一步**：統帥授權後執行 0a（110 /etc/hosts）+ 0b（188 cadvisor restart），觀察 30 分鐘後進 Phase 1 Migration。
diff --git a/docs/superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md b/docs/superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md
new file mode 100644
index 00000000..80459c4d
--- /dev/null
+++ b/docs/superpowers/specs/2026-04-18-blindspot-governance-capacity-l4.md
@@ -0,0 +1,890 @@
+# AWOOOI 監控盲區治理 + 容量自主化戰役總結（2026-04-18）
+
+> 🔴🔴🔴 **Single Source of Truth** — 本次 Session 所有決議、Schema、實施步驟的唯一事實來源
+> **啟動日**：2026-04-18 台北時間
+> **狀態**：🟡 Schema 定稿待 migration；Phase 0a/0b 止血指令待統帥授權執行
+> **上游藍圖**：[`2026-04-15-MASTER-ai-autonomous-flywheel-v2.md`](./2026-04-15-MASTER-ai-autonomous-flywheel-v2.md)（Phase 0-6 已完成）
+> **本檔定位**：Phase 7 — 盲區矯正 + 容量自主化（MASTER 未涵蓋的新層次）
+
+---
+
+## §0 TL;DR（一分鐘版）
+
+過去 13 天，110 主機 load=18.35、188 主機 load=16.32、cadvisor 狂吃 288% CPU —— **全部無人察覺**。原因不是系統沒壞，是 AWOOOI **監控覆蓋不到三成**（Prometheus 僅 35 targets / 58 rules），且**告警規則量錯維度**（CPU idle% 而非 load average）。統帥下令：**徹底重建監控地基 + 資料永久化 DB + AI 四分工（OpenClaw × NemoTron × Hermes × LLM）接管飛輪**。本檔交付：
+
+1. **10 張表資料庫架構**（Phase 1 基底）
+2. **四層防禦戰略**（Phase 3-6 施工藍圖）
+3. **7 個實施 Phase 的詳盡步驟**（從止血到 AI 自主治理）
+4. **與 7 項自動化引擎的完整對映**（每格誰負責）
+
+---
+
+## §1 觸發事件：MoWoooWorkDown 假警報
+
+```
+2026-04-18 凌晨
+  Blackbox probe mo.wooo.work → 超時 5s → Alertmanager 噴 P1
+  Telegram 推告警 → 統帥起床救火
+  curl https://mo.wooo.work  → 10s 回應但成功
+  → 應用層沒死，是探針死
+    ↓
+    深潛 RCA
+    ↓
+  真凶鎖定：
+    - 110 主機 load=18.35（非應用問題）
+    - ClickHouse 134% CPU
+    - DNS 解析 mo.wooo.work → UDP 封包在高負載下丟失
+    - Blackbox timeout 5s 太短 → 吃不下 DNS 4s + 應用 2s
+```
+
+**已修部分**：Blackbox timeout 5s→10s（commit 待記錄於 Gitea）
+
+**未解的根本**：
+- 為何 110 load=18.35 撐 13 天無人知？
+- 為何只有 mo.wooo.work 觸發，其他網站沒事？
+- 為何「沒告警」就以為系統健康？
+
+---
+
+## §2 鐵證數據（2026-04-18 12:00 台北實測）
+
+### 2.1 110 主機
+
+| 指標 | 數值 | 診斷 |
+|-----|------|-----|
+| uptime | 13 天 | — |
+| load5 | 18.35 | 🔴 持續飽和 13 天 |
+| ClickHouse CPU | 134% | 🔴 吞掉一個多核心 |
+| GitHub Actions Runner | 92% | 🟡 |
+| Kafka | 45% | 🟡 |
+| Sentry 容器數 | 40+ | 🔴 單一產品佔主機最大比例 |
+| Harbor 組件 | 5 個 | 🟡 |
+| Bitan / Gitea / Redis / Prometheus / Blackbox | 各 1-15% | 💢 全疊在一台 |
+
+### 2.2 188 主機
+
+| 指標 | 數值 | 診斷 |
+|-----|------|-----|
+| uptime | 13 天 | **與 110 同一時段開始過載** |
+| load5 | 14.19 / 15.91 / 16.32 | 🔴 12 核 → per-core 1.18，飽和 |
+| Memory | 62G / 8.4G used | ✅ 記憶體充裕 |
+| **cadvisor CPU** | **288%** | 🔴🔴🔴 **監控工具自爆 13 天無人知** |
+| clickhouse-server CPU | 169% | 🔴 SignOz 後端 |
+| prisma migrate deploy | 109% | 🟡 momo 短暫 |
+| postgres momo_analytics idle | 76% + 8.8% | 🟡 可疑連線洩漏 |
+
+### 2.3 Prometheus 覆蓋實況
+
+| 資源 | 數量 | 覆蓋評價 |
+|-----|------|---------|
+| active targets 總數 | **35** | 🔴 遠低於實際資產 |
+| alert rules 總數 | **58 條 / 16 組** | 🔴 覆蓋不全 |
+| firing rules | 2（HostHighCpuLoad + HostBackupFailed） | — |
+| node-exporter | 只監 110 / 112 / 188 | 🔴🔴🔴 **K3s Master 120/121 + VIP 125 完全沒監控** |
+| blackbox-http | 8 個網站 | 🔴 實際應有 15+ |
+| cadvisor | 只 cadvisor-110 | 🔴 188 無 cadvisor target |
+| kube-state-metrics | 部署了但 Prometheus 沒抓 | 🔴 K8s pod 狀態無告警基礎 |
+| Kafka exporter | 不存在 | 🔴 消息佇列完全黑箱 |
+| Ollama GPU / DCGM | 不存在 | 🔴 VRAM 黑箱 |
+| SSL cert expiry | 只 1 條 rule | 🟡 |
+
+### 2.4 規則品質問題
+
+**HostHighCpuLoad 規則**：
+```yaml
+expr: 100 - (avg by (host) (rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100) > 80
+for: 300s
+```
+
+**致命瑕疵**：測 CPU idle%，不是 load average。
+- 110 實測：load=18.35 時 CPU idle 可能還有 30% → 規則不觸發
+- 實際問題是**進程 queue 過長 + iowait 高 + UDP 丟包**，CPU 使用率不一定飆
+- **結果**：13 天過載零告警
+
+---
+
+## §3 根因盤點：為何結構性失守
+
+| # | 問題 | 本質 |
+|---|------|-----|
+| 1 | 監控量錯維度 | 用 `CPU idle` 不是 `load avg / iowait / run queue` |
+| 2 | 監控覆蓋不到三成 | 35 targets vs 實際數百個資產 |
+| 3 | 沒有容量配額 | Docker / K8s workload 大半無 `limits` |
+| 4 | 主機角色混雜 | 110/188 各疊 10+ 異質服務，無隔離 |
+| 5 | 新服務上線無容量評估 | 東西一直疊，沒人問「承載力夠嗎」 |
+| 6 | 監控工具自己沒被監控 | cadvisor 288% 活證據 |
+| 7 | 沒有趨勢預測 / AI 容量巡檢 | Phase 4 的 `trend_predictor` 建好但沒對主機跑 |
+
+**深層原因**：**寫了一堆 MD / ADR / Memory / Skill，但執行偏移**。MD 沒有強制力，不會阻止人把 40 個 Sentry 容器塞進 110。**需要把治理寫進 DB，讓 AI 強制檢查**。
+
+---
+
+## §4 與 AI 自主化北極星對齊
+
+### 4.1 七大自動化引擎（統帥明示）
+
+| # | 引擎 | 定義 | 當前狀態 |
+|---|-----|-----|---------|
+| E1 | **自動監控** | 從 Prometheus 時序動態感知基準線，自動發現異常 | 🔴 靜態閾值，規則量錯維度 |
+| E2 | **自動告警** | AI 分類 / 嚴重度 / 路由 / 抑制 | 🟡 Phase 2 Agent 協作已建，但 general 兜底 41% |
+| E3 | **自動建立匹配規則** | 規則由 AI 從歷史 Incident 學習產出 | 🔴 25 條全硬編 |
+| E4 | **自動匹配規則** | 來告警時 AI 判斷用哪條規則 | 🟡 Jaccard 相似度存在但低質 |
+| E5 | **自動建立 Playbook** | 修復劇本由 AI 動態生成 | 🟡 Phase 1 Solver K8s 清單注入，但仍幻覺 |
+| E6 | **自動修復** | AI 執行 + 驗證 + 失敗自我修正（stderr 回灌） | 🔴 approval_execution.py 不會回灌 stderr |
+| E7 | **自動建立 KM** | 故障完自動沉澱 RCA 進 KM + embedding | 🟡 KM +5/24h，目標 >20 |
+
+### 4.2 AI 四分工責任矩陣（LLM × OpenClaw × NemoTron × Hermes）
+
+| 引擎 | 主責 Agent | 輔責 Agent | 備註 |
+|-----|----------|----------|------|
+| E1 監控 | **NemoTron** 每日容量巡檢 | Hermes 趨勢分析 | NemoTron 8B 夠力做統計預測 |
+| E2 告警 | **OpenClaw** 分類分診 | Critic 挑戰 | 語意分類取代前綴硬編 |
+| E3 規則生成 | **Hermes** 分析師（RLHF 角色） | NemoTron 提案 | Hermes 擅長 reasoning + risk |
+| E4 規則匹配 | **OpenClaw** embedding 檢索 | NemoTron 排序 | Jaccard → semantic |
+| E5 Playbook 生成 | **NemoTron + Solver** | Reviewer 安全官 | 動態 K8s 清單注入 + dry-run |
+| E6 自動修復 | **Coordinator** 協調 | **Critic 挑戰** + **Verifier 驗證** | stderr 回灌機制 |
+| E7 KM 沉澱 | **Hermes** 分析敘事 | OpenClaw 摘要 | Hermes reasoning → 敘事卡片 |
+
+**鐵律**：每個 Agent 動作**必須**寫入 `ai_collaboration_trace`，包括 prompt / response / confidence / challenged_by。
+
+---
+
+## §5 資料庫地基：10 張表完整 Schema
+
+### 5.1 架構圖
+
+```
+asset_discovery_run ─┬─> asset_inventory ─┬─> asset_relationship (dependency graph)
+                     │    │               ├─> asset_coverage_snapshot (resource × 7 engines)
+                     │    │               ├─> asset_compliance_snapshot (SSL/CVE/secret/backup)
+                     │    │               └─> asset_change_event (diff between runs)
+                     │    │
+                     │    └─> alert_rule_catalog (rules as first-class assets)
+                     │
+                     ├─> host_capacity_snapshot (每日主機容量快照)
+                     │
+                     ├─> capacity_violation_event (配額違規記錄)
+                     │
+                     └─> automation_operation_log (every AI op audit trail)
+                              │
+                              └─> ai_collaboration_trace (multi-agent handoff)
+```
+
+### 5.2 完整 DDL
+
+```sql
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 1: asset_inventory — 全景資產清單（核心主表）
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_inventory (
+    asset_id          BIGSERIAL PRIMARY KEY,
+    asset_key         TEXT NOT NULL UNIQUE,          -- 穩定識別：host:110 / container:110:sentry-web / k8s:awoooi-prod/awoooi-api
+    asset_type        TEXT NOT NULL,
+    parent_asset_id   BIGINT REFERENCES asset_inventory(asset_id),
+    environment       TEXT NOT NULL DEFAULT 'prod',  -- prod / staging / dev（advisor 建議）
+    host              TEXT,                           -- 110 / 120 / 121 / 188 / 112 / 125
+    namespace         TEXT,
+    name              TEXT NOT NULL,
+    metadata          JSONB NOT NULL DEFAULT '{}'::jsonb,
+    tags              TEXT[] NOT NULL DEFAULT '{}',
+    owner_team        TEXT,
+    criticality       TEXT CHECK (criticality IN ('P0','P1','P2','P3')),
+    data_classification TEXT CHECK (data_classification IN ('public','internal','sensitive','secret')),
+    external          BOOLEAN NOT NULL DEFAULT false,  -- 第三方 SaaS (Cloudflare/NVIDIA/Gemini)
+    lifecycle_state   TEXT NOT NULL DEFAULT 'active'
+                      CHECK (lifecycle_state IN ('planned','provisioning','active','degraded','deprecated','decommissioned')),
+    source_repo       TEXT,                           -- Git repo URL
+    source_commit_sha TEXT,                           -- 部署的 commit
+    -- 容量欄位（Layer 4 AI 巡檢用）
+    cpu_avg_7d        NUMERIC(5,2),
+    mem_avg_7d        NUMERIC(5,2),
+    capacity_headroom NUMERIC(5,2),                   -- 剩餘承載 0.00-1.00
+    resource_limits   JSONB,                          -- {cpu:"2",mem:"4Gi"}
+    resource_requests JSONB,
+    quota_violation_count INT NOT NULL DEFAULT 0,
+    sla_target        JSONB,                          -- {uptime:99.9, rto_min:15, rpo_min:5}
+    cost_monthly_usd  NUMERIC(10,2),
+    first_seen_at     TIMESTAMPTZ NOT NULL DEFAULT now(),
+    last_seen_at      TIMESTAMPTZ NOT NULL DEFAULT now(),
+    decommissioned_at TIMESTAMPTZ,
+    created_at        TIMESTAMPTZ NOT NULL DEFAULT now(),
+    updated_at        TIMESTAMPTZ NOT NULL DEFAULT now(),
+    CONSTRAINT asset_type_valid CHECK (asset_type IN (
+        'host','container','k8s_workload','k8s_resource','database','table',
+        'website','api_endpoint','package','log_stream','km_entry',
+        'frontend','backend','ci_pipeline','gitea_repo','monitoring_target',
+        'secret','volume','network','certificate','scheduled_job',
+        'message_queue','cache','dashboard','ai_agent','llm_model',
+        'third_party_service','backup_target'
+    ))
+);
+CREATE INDEX idx_ai_type_host ON asset_inventory(asset_type, host);
+CREATE INDEX idx_ai_env ON asset_inventory(environment, lifecycle_state);
+CREATE INDEX idx_ai_metadata ON asset_inventory USING GIN (metadata);
+CREATE INDEX idx_ai_tags ON asset_inventory USING GIN (tags);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 2: asset_discovery_run — 每次盤點 header
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_discovery_run (
+    run_id             UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    triggered_by       TEXT NOT NULL,                -- human:ogt / cron:daily / incident:INC-123 / ai:proactive_inspector
+    scope              TEXT[] NOT NULL,              -- 本次掃描的 asset_type 清單
+    scan_depth         TEXT NOT NULL DEFAULT 'shallow'
+                       CHECK (scan_depth IN ('shallow','deep','full')),
+    host_filter        TEXT[],                       -- 本次只掃哪些主機
+    started_at         TIMESTAMPTZ NOT NULL DEFAULT now(),
+    ended_at           TIMESTAMPTZ,
+    status             TEXT NOT NULL CHECK (status IN ('running','success','partial','failed','aborted')),
+    total_assets       INT,
+    new_assets         INT NOT NULL DEFAULT 0,
+    modified_assets    INT NOT NULL DEFAULT 0,
+    disappeared_assets INT NOT NULL DEFAULT 0,
+    tools_used         JSONB,                        -- {"ssh":5,"kubectl":1,"prom_api":1,"gitea_api":1}
+    duration_ms        INT,
+    error              TEXT,
+    summary            JSONB
+);
+CREATE INDEX idx_adr_started ON asset_discovery_run(started_at DESC);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 3: asset_coverage_snapshot — 資產 × 7 自動化 覆蓋矩陣
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_coverage_snapshot (
+    snapshot_id       BIGSERIAL PRIMARY KEY,
+    run_id            UUID NOT NULL REFERENCES asset_discovery_run(run_id) ON DELETE CASCADE,
+    asset_id          BIGINT NOT NULL REFERENCES asset_inventory(asset_id),
+    dimension         TEXT NOT NULL CHECK (dimension IN (
+        'auto_monitoring','auto_alerting','auto_rule_creation',
+        'auto_rule_matching','auto_playbook','auto_remediation','auto_km_creation'
+    )),
+    coverage_status   TEXT NOT NULL CHECK (coverage_status IN ('green','yellow','red','unknown')),
+    evidence          JSONB NOT NULL DEFAULT '{}'::jsonb,  -- {prom_target,rule_name,playbook_id,km_entry_id}
+    gap_reason        TEXT,
+    recommended_action TEXT,
+    confidence        NUMERIC(3,2),
+    detected_by       TEXT NOT NULL,                 -- scanner:blackbox / ai:nemotron / human
+    created_at        TIMESTAMPTZ NOT NULL DEFAULT now(),
+    UNIQUE (run_id, asset_id, dimension)
+);
+CREATE INDEX idx_acs_asset ON asset_coverage_snapshot(asset_id, dimension);
+CREATE INDEX idx_acs_red_yellow ON asset_coverage_snapshot(coverage_status) WHERE coverage_status IN ('red','yellow');
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 4: asset_relationship — 資產依賴圖（爆炸半徑必需）
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_relationship (
+    relationship_id   BIGSERIAL PRIMARY KEY,
+    from_asset_id     BIGINT NOT NULL REFERENCES asset_inventory(asset_id),
+    to_asset_id       BIGINT NOT NULL REFERENCES asset_inventory(asset_id),
+    relationship_type TEXT NOT NULL CHECK (relationship_type IN (
+        'depends_on','calls','stores_data_in','backs_up_to',
+        'routes_to','authenticates_via','monitors','alerts_to','logs_to'
+    )),
+    strength          NUMERIC(3,2),                  -- 0.00-1.00 依賴強度
+    metadata          JSONB,
+    first_detected_at TIMESTAMPTZ NOT NULL DEFAULT now(),
+    last_verified_at  TIMESTAMPTZ,
+    is_active         BOOLEAN NOT NULL DEFAULT true,
+    UNIQUE (from_asset_id, to_asset_id, relationship_type)
+);
+CREATE INDEX idx_ar_from ON asset_relationship(from_asset_id);
+CREATE INDEX idx_ar_to ON asset_relationship(to_asset_id);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 5: alert_rule_catalog — 告警規則本身即資產
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE alert_rule_catalog (
+    rule_id           BIGSERIAL PRIMARY KEY,
+    rule_name         TEXT NOT NULL UNIQUE,
+    source            TEXT NOT NULL CHECK (source IN ('yaml_hardcoded','ai_generated','human_written','playbook_derived')),
+    expr              TEXT NOT NULL,
+    duration_seconds  INT,
+    severity          TEXT,
+    labels            JSONB,
+    annotations       JSONB,
+    linked_asset_ids  BIGINT[],                      -- 這條規則保護的資產
+    created_by_agent  TEXT,                          -- nemotron / hermes / openclaw / human:ogt
+    -- 規則品質追蹤
+    true_positive_count  INT NOT NULL DEFAULT 0,
+    false_positive_count INT NOT NULL DEFAULT 0,
+    noise_rate           NUMERIC(5,2),
+    last_fired_at        TIMESTAMPTZ,
+    -- 信心與演化
+    confidence        NUMERIC(3,2),
+    review_status     TEXT CHECK (review_status IN ('draft','approved','deprecated','retired')),
+    superseded_by_rule_id BIGINT REFERENCES alert_rule_catalog(rule_id),
+    created_at        TIMESTAMPTZ NOT NULL DEFAULT now(),
+    updated_at        TIMESTAMPTZ NOT NULL DEFAULT now()
+);
+CREATE INDEX idx_arc_source ON alert_rule_catalog(source);
+CREATE INDEX idx_arc_assets ON alert_rule_catalog USING GIN (linked_asset_ids);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 6: asset_change_event — 資產變化追蹤
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_change_event (
+    event_id          BIGSERIAL PRIMARY KEY,
+    run_id            UUID NOT NULL REFERENCES asset_discovery_run(run_id),
+    asset_id          BIGINT REFERENCES asset_inventory(asset_id),
+    change_type       TEXT NOT NULL CHECK (change_type IN (
+        'asset_added','asset_removed','asset_modified',
+        'coverage_improved','coverage_degraded',
+        'criticality_changed','owner_changed','lifecycle_changed'
+    )),
+    before_state      JSONB,
+    after_state       JSONB,
+    diff              JSONB,
+    detected_at       TIMESTAMPTZ NOT NULL DEFAULT now(),
+    ai_analysis       TEXT
+);
+CREATE INDEX idx_ace_run ON asset_change_event(run_id);
+CREATE INDEX idx_ace_asset ON asset_change_event(asset_id, detected_at DESC);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 7: asset_compliance_snapshot — 合規狀態快照
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE asset_compliance_snapshot (
+    snapshot_id       BIGSERIAL PRIMARY KEY,
+    run_id            UUID REFERENCES asset_discovery_run(run_id),
+    asset_id          BIGINT NOT NULL REFERENCES asset_inventory(asset_id),
+    dimension         TEXT NOT NULL CHECK (dimension IN (
+        'ssl_cert_valid','cve_scan','secret_rotated','backup_tested',
+        'audit_log_enabled','access_reviewed','encryption_at_rest'
+    )),
+    status            TEXT NOT NULL CHECK (status IN ('compliant','warning','violation','unknown')),
+    expires_at        TIMESTAMPTZ,                   -- cert 到期 / 下次掃描
+    detail            JSONB,
+    remediation_deadline TIMESTAMPTZ,
+    detected_at       TIMESTAMPTZ NOT NULL DEFAULT now(),
+    UNIQUE (asset_id, dimension, snapshot_id)
+);
+CREATE INDEX idx_comp_asset ON asset_compliance_snapshot(asset_id, dimension);
+CREATE INDEX idx_comp_expiring ON asset_compliance_snapshot(expires_at) WHERE expires_at IS NOT NULL;
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 8: host_capacity_snapshot — 主機容量快照（每日 by NemoTron）
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE host_capacity_snapshot (
+    snapshot_id       BIGSERIAL PRIMARY KEY,
+    host              TEXT NOT NULL,
+    captured_at       TIMESTAMPTZ NOT NULL DEFAULT now(),
+    load1             NUMERIC(6,2),
+    load5             NUMERIC(6,2),
+    load15            NUMERIC(6,2),
+    cpu_used_pct      NUMERIC(5,2),
+    cpu_iowait_pct    NUMERIC(5,2),
+    mem_used_pct      NUMERIC(5,2),
+    swap_used_pct     NUMERIC(5,2),
+    disk_used_pct     JSONB,                         -- per mount
+    container_count   INT,
+    k8s_pod_count     INT,
+    top_cpu_offenders JSONB,                         -- top 5 進程
+    top_mem_offenders JSONB,
+    headroom_pct      NUMERIC(5,2),                  -- AI 估算剩餘承載
+    ai_verdict        TEXT,                          -- NemoTron 判斷：safe/warning/critical
+    ai_reasoning      TEXT,
+    recommended_actions JSONB,
+    written_by_agent  TEXT NOT NULL                  -- nemotron / human:ogt
+);
+CREATE INDEX idx_hcs_host_time ON host_capacity_snapshot(host, captured_at DESC);
+CREATE INDEX idx_hcs_critical ON host_capacity_snapshot(ai_verdict) WHERE ai_verdict IN ('warning','critical');
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 9: capacity_violation_event — 容量配額違規
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE capacity_violation_event (
+    event_id          BIGSERIAL PRIMARY KEY,
+    asset_id          BIGINT REFERENCES asset_inventory(asset_id),
+    host              TEXT,
+    violation_type    TEXT NOT NULL CHECK (violation_type IN (
+        'no_limit_set','over_request','over_limit','host_saturation',
+        'over_sla_budget','unauthorized_new_deploy'
+    )),
+    threshold         NUMERIC(10,2),
+    actual_value      NUMERIC(10,2),
+    detected_at       TIMESTAMPTZ NOT NULL DEFAULT now(),
+    auto_action       TEXT,                          -- AI 採取的動作
+    auto_action_op_id UUID,                          -- 對應 automation_operation_log
+    human_override    TEXT,
+    resolved_at       TIMESTAMPTZ
+);
+CREATE INDEX idx_cve_asset ON capacity_violation_event(asset_id, detected_at DESC);
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 10: automation_operation_log — 所有 AI 自動化動作稽核主表 🔴
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE automation_operation_log (
+    op_id             UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    operation_type    TEXT NOT NULL CHECK (operation_type IN (
+        'monitor_configured','monitor_removed',
+        'alert_fired','alert_suppressed','alert_routed',
+        'rule_created','rule_updated','rule_matched','rule_rejected','rule_deprecated',
+        'playbook_generated','playbook_updated','playbook_executed',
+        'remediation_executed','remediation_verified','remediation_rolled_back',
+        'self_correction_attempted',                  -- 引擎 4 stderr 回灌
+        'km_created','km_updated','km_linked',
+        'asset_discovered','coverage_recalculated',
+        'capacity_recommendation','quota_enforced'
+    )),
+    asset_id          BIGINT REFERENCES asset_inventory(asset_id),
+    incident_id       BIGINT,
+    run_id            UUID REFERENCES asset_discovery_run(run_id),
+    actor             TEXT NOT NULL,                 -- openclaw / nemotron / hermes / solver / critic / coordinator / claude-code / human:ogt
+    input             JSONB NOT NULL DEFAULT '{}'::jsonb,
+    output            JSONB NOT NULL DEFAULT '{}'::jsonb,
+    dry_run_result    JSONB,                         -- dry-run 預覽結果
+    status            TEXT NOT NULL CHECK (status IN ('pending','success','failed','dry_run','rolled_back')),
+    error             TEXT,
+    duration_ms       INT,
+    tokens_in         INT,
+    tokens_out        INT,
+    cost_usd          NUMERIC(10,6),
+    budget_bucket     TEXT,                          -- 分攤到哪個預算桶
+    parent_op_id      UUID REFERENCES automation_operation_log(op_id),
+    retry_count       INT NOT NULL DEFAULT 0,        -- advisor 建議
+    retry_of_op_id    UUID REFERENCES automation_operation_log(op_id),
+    stderr_feed_back  TEXT,                          -- 引擎 4：上次 stderr 回灌 LLM 的內容
+    tags              TEXT[],                        -- technical_debt / blind_spot_mitigation / ...
+    created_at        TIMESTAMPTZ NOT NULL DEFAULT now()
+);
+CREATE INDEX idx_aol_type_time ON automation_operation_log(operation_type, created_at DESC);
+CREATE INDEX idx_aol_asset ON automation_operation_log(asset_id, created_at DESC);
+CREATE INDEX idx_aol_incident ON automation_operation_log(incident_id) WHERE incident_id IS NOT NULL;
+CREATE INDEX idx_aol_actor ON automation_operation_log(actor, created_at DESC);
+CREATE INDEX idx_aol_retry ON automation_operation_log(retry_of_op_id) WHERE retry_of_op_id IS NOT NULL;
+
+-- ═══════════════════════════════════════════════════════════════════
+-- TABLE 11: ai_collaboration_trace — 多 Agent 協作逐步（LLM × OpenClaw × NemoTron × Hermes）
+-- ═══════════════════════════════════════════════════════════════════
+CREATE TABLE ai_collaboration_trace (
+    trace_id          BIGSERIAL PRIMARY KEY,
+    op_id             UUID NOT NULL REFERENCES automation_operation_log(op_id) ON DELETE CASCADE,
+    step_order        INT NOT NULL,
+    agent             TEXT NOT NULL,                 -- openclaw / nemotron / hermes / diagnostician / solver / reviewer / critic / coordinator / executor / verifier
+    model             TEXT,                          -- nemotron-70b / hermes-3 / openclaw-v1 / claude-opus-4-7
+    system_prompt_version TEXT,
+    prompt            TEXT,
+    response          JSONB,
+    confidence        NUMERIC(3,2),
+    challenged_by     TEXT[],
+    accepted          BOOLEAN,
+    tokens_in         INT,
+    tokens_out        INT,
+    duration_ms       INT,
+    created_at        TIMESTAMPTZ NOT NULL DEFAULT now(),
+    UNIQUE (op_id, step_order)
+);
+CREATE INDEX idx_act_op ON ai_collaboration_trace(op_id, step_order);
+CREATE INDEX idx_act_agent ON ai_collaboration_trace(agent, created_at DESC);
+```
+
+> 註：雖然架構圖標題說 10 張表，完整 DDL 含 `ai_collaboration_trace` 共 **11 張表**（原 6 + asset_relationship + alert_rule_catalog + asset_compliance_snapshot + host_capacity_snapshot + capacity_violation_event）。
+
+### 5.3 與現有表關係
+
+| 現有表 | 連動方式 |
+|-------|---------|
+| `incidents` | `automation_operation_log.incident_id` FK |
+| `playbooks` | `asset_coverage_snapshot.evidence->>'playbook_id'` |
+| `knowledge_entries` | `asset_coverage_snapshot.evidence->>'km_entry_id'` |
+| `timeline_events` | 每次 `automation_operation_log` 寫入同步追加 event |
+| `alert_rules.yaml` | seed → `alert_rule_catalog` (source='yaml_hardcoded')，未來漸進 AI 產出 |
+
+### 5.4 保留策略
+
+| 表 | 保留期 | 清理 |
+|---|-------|-----|
+| asset_inventory | 永久 | soft delete |
+| asset_discovery_run | 永久 | — |
+| asset_coverage_snapshot | 90 天 | 保最新 + 每週一筆 |
+| asset_relationship | 永久 | is_active=false |
+| alert_rule_catalog | 永久 | review_status=retired |
+| asset_change_event | 永久 | — |
+| asset_compliance_snapshot | 180 天 | — |
+| host_capacity_snapshot | 90 天 | — |
+| capacity_violation_event | 永久 | — |
+| automation_operation_log | 永久 | 未來 archive ClickHouse |
+| ai_collaboration_trace | 180 天 | 低信心 / rejected 永久保留作訓料 |
+
+---
+
+## §6 四層防禦策略（長期根治）
+
+### Layer 1：告警規則精準化（Phase 3，本週）
+
+**新增 / 修正規則**（寫進 `alert_rules.yaml` + seed 進 `alert_rule_catalog`）：
+
+```yaml
+# === 取代現有 HostHighCpuLoad（量錯維度） ===
+- alert: HostLoadPersistentHigh
+  expr: node_load5 / count by(host) (node_cpu_seconds_total{mode="idle"}) > 0.9
+  for: 30m
+  labels: { severity: warning, auto_repair: "false" }
+  annotations: { summary: "{{$labels.host}} load/core 持續 30 分鐘 > 0.9" }
+
+- alert: HostLoadCritical
+  expr: node_load5 / count by(host) (node_cpu_seconds_total{mode="idle"}) > 1.5
+  for: 10m
+  labels: { severity: critical }
+
+- alert: HostIowaitHigh
+  expr: avg by(host) (rate(node_cpu_seconds_total{mode="iowait"}[5m])) > 0.2
+  for: 10m
+
+# === 監控工具自我監控（cadvisor 真實案例） ===
+- alert: CadvisorSelfCpuHigh
+  expr: rate(container_cpu_usage_seconds_total{name="cadvisor"}[5m]) > 1.0
+  for: 5m
+  annotations: { summary: "cadvisor 自己吃超過 1 個核心" }
+
+- alert: PrometheusSelfScrapeFailed
+  expr: up{job="prometheus"} == 0
+  for: 1m
+
+- alert: AlertmanagerSelfDown
+  expr: up{job="alertmanager"} == 0
+  for: 1m
+
+# === K3s 補監控 ===
+- alert: K3sControlPlaneDown
+  expr: up{job="node-exporter",host=~"120|121"} == 0
+  for: 2m
+  labels: { severity: critical }
+
+- alert: K3sEtcdLagHigh
+  expr: etcd_mvcc_db_total_size_in_bytes > 4e9
+  for: 10m
+
+# === 容器配額違規 ===
+- alert: ContainerNoMemoryLimit
+  expr: count by(container) (container_spec_memory_limit_bytes == 0) > 0
+  for: 1h
+  annotations: { summary: "容器 {{$labels.container}} 無 memory limit" }
+
+# === 趨勢預測（Phase 4 接線） ===
+- alert: HostCapacityTrendCritical
+  expr: predict_linear(node_load5[1h], 7*24*3600) > count by(host)(node_cpu_seconds_total{mode="idle"})
+  for: 30m
+  annotations: { summary: "{{$labels.host}} 預測 7 天內撞飽和" }
+```
+
+### Layer 2：強制資源配額（Phase 5，本月）
+
+**執行項**：
+
+| # | 動作 | 負責人 | 驗收 |
+|---|-----|-------|------|
+| L2.1 | 審 110/188 所有 `docker-compose.yml`，補 `deploy.resources.limits` | 人工 + AI 輔助 | 所有 compose 有 limit |
+| L2.2 | K3s 每 namespace 建 `ResourceQuota` + `LimitRange` | DevOps | `kubectl describe quota -A` 全覆蓋 |
+| L2.3 | Gitea CI 加檢查：缺 limit → PR reject | 新增 workflow | blocked PR 次數可見 |
+| L2.4 | Harbor / Gitea / Sentry / ClickHouse 強制上限（重型） | 人工排期 | `docker inspect` 看 limit 不為 0 |
+| L2.5 | 違規記入 `capacity_violation_event` | 自動 | 每日有新記錄 |
+
+### Layer 3：主機角色重分工（Phase 6，本季）
+
+**現況 → 目標**：
+
+| 主機 | 當前 | 目標角色 | 搬遷對象 |
+|-----|------|---------|---------|
+| 110 | DevOps + 監控 + CI + Sentry + Kafka + 業務 (Bitan) | **DevOps 金庫**：Harbor / Gitea / CI runner | Sentry / Kafka / Bitan 搬走 |
+| 120 / 121 | K3s Master | **K3s Control-Plane（純）** | 無（維持） |
+| 188 | K3s PG + Ollama + SignOz + momo + 18 容器 | **AI + 資料層**：PG / Ollama / Redis / momo | SignOz ClickHouse / Grafana / Loki 搬走 |
+| **NEW-A** | — | **監控專用主機** | Prometheus + Alertmanager + Grafana + Loki + SignOz |
+| **NEW-B** | — | **冷儲存 + 重型業務** | Sentry / MinIO / Kafka / 備份目標 |
+
+**搬遷順序**（降風險）：
+1. 先建 NEW-A（監控獨立）→ 解除 110/188 觀測性耦合
+2. 搬 Sentry 從 110 → NEW-B
+3. 搬 SignOz ClickHouse 從 188 → NEW-A
+4. 保留 110 專做 DevOps Vault
+
+### Layer 4：AI 自主治理（Phase 4，2 週）
+
+**NemoTron 每日容量巡檢（MVP）**：
+
+```
+cron: 每日 02:00 台北
+↓
+NemoTron 讀取：
+  - 過去 7 天 Prometheus metrics (load5, CPU, mem, iowait per host)
+  - asset_inventory (每資產 expected resources)
+  - host_capacity_snapshot (歷史基線)
+↓
+Hermes 趨勢分析：
+  - predict_linear(load5[7d], 7d) → 未來 7 天預測
+  - 對比 asset_inventory.sla_target
+  - 標出「撞牆風險」資產
+↓
+OpenClaw 提案生成：
+  - 建議搬遷計畫（低優先容器 → 低負載主機）
+  - 建議新增 resource limit
+  - 建議廢棄長期低使用容器
+↓
+寫入：
+  - host_capacity_snapshot (N 筆)
+  - automation_operation_log (operation_type=capacity_recommendation)
+  - ai_collaboration_trace (每步)
+↓
+Telegram 早報：Top 3 風險 + Top 3 建議 → 統帥 Approve → 執行 or Reject
+```
+
+---
+
+## §7 實施階段（按順序，風險遞增）
+
+### Phase 0a：110 DNS 止血（10 分鐘）
+
+```bash
+# 1. 靜態綁定 mo.wooo.work → 避免 DNS 超時
+ssh 192.168.0.110 "echo '114.32.151.246 mo.wooo.work' | sudo tee -a /etc/hosts"
+
+# 2. 驗證
+ssh 192.168.0.110 "getent hosts mo.wooo.work"
+
+# 3. 等 3 分鐘 → 觀察 load 變化
+ssh 192.168.0.110 "uptime"
+```
+
+**風險**：🟡 中（IP 寫死是毒蘋果，Phase 6 搬監控時必須廢除）
+**回滾**：`sudo sed -i '/mo.wooo.work/d' /etc/hosts`
+**稽核**：寫入 `automation_operation_log`（tags=['technical_debt','blind_spot_mitigation']）
+
+### Phase 0b：188 cadvisor 降壓（5 分鐘）
+
+```bash
+# 1. 重啟 cadvisor（無狀態，釋放 288% CPU）
+ssh ollama@192.168.0.188 "docker restart cadvisor"
+
+# 2. 等 3 分鐘 → 驗證
+ssh ollama@192.168.0.188 "docker stats --no-stream cadvisor | tail -1"
+
+# 3. 檢查 188 load
+ssh ollama@192.168.0.188 "uptime"
+```
+
+**風險**：🟢 低（cadvisor 無狀態）
+**回滾**：`docker start cadvisor`（但 restart 本身即自愈）
+**稽核**：寫入 `automation_operation_log`（tags=['blind_spot_mitigation']）
+**後續追查**：cadvisor 為何燒 288% 13 天？可能版本 bug 或配置錯誤，歸入 Phase 2 根因追溯
+
+### Phase 0c：Prometheus blackbox /health 改造（15 分鐘，選）
+
+```yaml
+# prometheus.yml 修改
+- job_name: 'blackbox-http'
+  static_configs:
+    - targets:
+        - https://mo.wooo.work/health   # 從 / 改成 /health
+```
+
+```bash
+ssh 192.168.0.110 "docker kill -s SIGHUP prometheus"
+```
+
+**風險**：🟢 極低
+**先決條件**：momo-pro-system 確認 /health 存在且 < 100ms
+
+### Phase 1：Alembic Migration 建 11 張表（30 分鐘）
+
+```bash
+# 位置
+cd /Users/ogt/awoooi/apps/api
+alembic revision -m "asset_inventory_foundation_v1" --autogenerate=false
+# 編輯新生成的 migration file，填入 §5.2 完整 DDL
+alembic upgrade head
+```
+
+**驗收**：
+```sql
+SELECT table_name FROM information_schema.tables
+WHERE table_schema='public' AND table_name LIKE 'asset_%' OR table_name LIKE 'automation_%' OR table_name LIKE 'host_capacity%';
+-- 應回傳 11 張表
+```
+
+**風險**：🟢 低（DDL only，PG 本身不在 188 top CPU）
+**回滾**：`alembic downgrade -1`
+
+### Phase 2：全景資產首掃（2 小時，限流）
+
+**執行腳本**：`apps/api/scripts/audit_assets_v1.py`
+
+```
+For host in [110, 120, 121, 188, 112, 125]:
+  scan:
+    - ssh: uptime / free / df / systemctl list-units --type=service --state=running
+    - docker ps / docker stats (with sleep 2s between hosts)
+    - kubectl get all,cm,secret,pvc -A (only on 120 via kubectl)
+    - Gitea API: repos / webhooks
+    - Prometheus API: targets / rules
+  write:
+    - asset_inventory (upsert by asset_key)
+    - asset_discovery_run (header)
+    - asset_coverage_snapshot (每 asset × 7 dimensions)
+    - asset_relationship (container→host, k8s→namespace)
+    - alert_rule_catalog (seed from alert_rules.yaml)
+```
+
+**限流**：每主機 sleep 2s 間隔，不平行打。
+
+### Phase 3：Layer 1 告警規則升級（1 小時）
+
+1. 寫入 `alert_rules.yaml`（§6 Layer 1 規則）
+2. Prometheus reload
+3. 每條新規則同步寫 `alert_rule_catalog`（source='yaml_hardcoded'，created_by_agent='claude-code'）
+4. 等 30 分鐘觀察實際觸發
+
+### Phase 4：NemoTron 容量巡檢 MVP（2 週）
+
+**新增服務**：`apps/api/src/services/capacity_inspector.py`
+
+```python
+class CapacityInspector:
+    async def daily_sweep(self):
+        for host in self.hosts:
+            metrics = await prometheus.query_range(...)
+            snapshot = HostCapacitySnapshot(host=host, ...)
+            verdict = await nemotron.analyze(metrics)    # 診斷
+            trend = await hermes.predict(metrics)        # 趨勢
+            recommendations = await openclaw.propose(snapshot, verdict, trend)
+            await db.save(snapshot, automation_op_log, ai_trace)
+            await telegram.push_daily_report(top_3_risks, top_3_proposals)
+```
+
+**cron**：K8s CronJob 每日 02:00 台北
+**驗收**：7 天後 `host_capacity_snapshot` 至少 35 筆（5 主機 × 7 天）
+
+### Phase 5：資源配額強制（Layer 2，1 個月）
+
+- 建 `.gitea/workflows/check_resource_limits.yml`
+- 批次補 compose limit
+- K3s 建 `ResourceQuota` per namespace
+- 違規寫 `capacity_violation_event`
+
+### Phase 6：主機搬遷（Layer 3，1 季）
+
+需採購決策，不在本 session 範圍。本檔僅登記計畫。
+
+### Phase 7：引擎 3+4 閉環強化（顧問已批准的引擎 4 stderr 回灌）
+
+- `approval_execution.py` retry loop 加 stderr 回灌邏輯
+- `post_execution_verifier.py` 接 `dynamic_baseline_service.py`
+- 每次自我修正寫 `automation_operation_log`（op_type='self_correction_attempted', stderr_feed_back=<stderr>）
+
+---
+
+## §8 HARD_RULES 新增（進 docs/HARD_RULES.md）
+
+1. **監控工具必須被監控** — cadvisor / Prometheus / Alertmanager / Grafana / Loki 自己的 CPU/Mem 必須在 `alert_rule_catalog` 有規則（cadvisor 288% 案例為鐵證）
+2. **配額即義務** — Docker compose 缺 `deploy.resources.limits` → Gitea CI PR reject
+3. **主機容量警戒線** — `host_capacity_snapshot.headroom_pct < 0.2` 三日 → 自動凍結新部署到該主機
+4. **告警維度鐵律** — 主機資源告警必須包含 `load_avg / iowait / run_queue`，不能只測 `CPU idle`
+5. **IP 靜態綁定技術債登記** — 任何 `/etc/hosts` 寫死必須寫 `automation_operation_log` tags='technical_debt' + `capacity_violation_event`
+6. **監控覆蓋率 SLO** — asset_coverage_snapshot 中 `red` 比例必須 < 20%（違反觸發 Telegram 日報警示）
+7. **新服務上線 Gate** — 任何新 K8s workload / Docker 容器上線前必須在 `asset_inventory` 有對應 entry + resource_limits 已填
+
+---
+
+## §9 AI 分工責任矩陣（11 張表 × 4 Agents）
+
+| 表 | OpenClaw | NemoTron | Hermes | LLM (Claude) |
+|---|---------|---------|--------|--------------|
+| asset_inventory | 分類 asset_type | 每日巡檢寫 capacity 欄位 | — | 新資產上線人工輔助 |
+| asset_discovery_run | 排程觸發 | — | — | 手動觸發 |
+| asset_coverage_snapshot | 計算覆蓋狀態 | 異常告警 | 趨勢分析 | — |
+| asset_relationship | 從告警 label 推依賴 | 主機拓撲 | — | — |
+| alert_rule_catalog | 規則匹配 (E4) | 規則建議 (E3) | 規則品質分析 (noise_rate) | 人工 review |
+| asset_change_event | 變化解讀 | — | 趨勢評估 | — |
+| asset_compliance_snapshot | SSL cert 掃描 | — | — | — |
+| host_capacity_snapshot | — | **主寫者**（E1） | 趨勢預測 | — |
+| capacity_violation_event | 自動偵測 | — | — | — |
+| automation_operation_log | 所有 Agent 都寫 | 所有 Agent 都寫 | 所有 Agent 都寫 | claude-code 寫 |
+| ai_collaboration_trace | step=diagnostician/solver/reviewer | step=capacity_inspector | step=analyst | step=coordinator |
+
+---
+
+## §10 驗收指標（量化 SLO）
+
+| 指標 | 當前 | 4 週目標 | 12 週目標 |
+|-----|------|---------|----------|
+| Prometheus active targets | 35 | 80+ | 150+ |
+| alert_rules 總數 | 58 | 90+ | 120+ |
+| asset_inventory 資產總數 | 0 | 200+ | 500+ |
+| asset_coverage_snapshot red 比例 | 未知 | < 40% | < 20% |
+| host_capacity_snapshot 筆數/日 | 0 | 5（一主機一筆） | 5 |
+| automation_operation_log 寫入/日 | 低 | 100+ | 500+ |
+| ai_collaboration_trace agent=nemotron 筆數/日 | 0 | 5+ | 30+ |
+| cadvisor 自身 CPU | 288% | < 50% | < 30% |
+| 110 load5 | 18.35 | < 8 | < 6 |
+| 188 load5 | 16.32 | < 8 | < 6 |
+| self_correction_attempted op 次數/週 | 0 | 5+ | 30+（表示 AI 會自我修正） |
+
+---
+
+## §11 技術債登記（與 Phase 0 綁定）
+
+| ID | 技術債 | 產生 Phase | 到期 Phase |
+|----|-------|----------|----------|
+| TD-001 | `/etc/hosts` 靜態綁定 mo.wooo.work | Phase 0a | Phase 6 搬監控時廢除 |
+| TD-002 | cadvisor 重啟但未根治 288% 原因 | Phase 0b | Phase 2 根因追溯 |
+| TD-003 | HostHighCpuLoad 規則保留（給 Phase 3 替換） | 本 session | Phase 3 新規則上線後 retire |
+| TD-004 | 110 主機超載根源未解（Sentry/Kafka/ClickHouse 疊一起） | 歷史 | Phase 6 搬遷 |
+| TD-005 | 188 主機超載根源未解 | 歷史 | Phase 6 搬遷 |
+| TD-006 | K3s 120/121/125 無 node-exporter | 歷史 | Phase 2 審計補 |
+| TD-007 | Kafka / Ollama GPU / ClickHouse 無 exporter | 歷史 | Phase 2 審計補 |
+
+**每條 TD 寫 `automation_operation_log` tags=['technical_debt','TD-XXX']**。
+
+---
+
+## §12 回滾計畫（每階段獨立）
+
+| Phase | 回滾方式 |
+|-------|---------|
+| 0a | `sed -i '/mo.wooo.work/d' /etc/hosts` |
+| 0b | 重啟失敗則 `docker start cadvisor`（cadvisor 無狀態） |
+| 0c | prometheus.yml 還原 + SIGHUP |
+| 1 | `alembic downgrade -1` |
+| 2 | audit 失敗不影響既有系統，只是 DB 資料不全 |
+| 3 | 移除新增 rule block + SIGHUP |
+| 4 | 停 CronJob，既有資料保留 |
+| 5 | 撤回 workflow 檢查 |
+| 6 | 搬遷分段，每段可獨立回滾 |
+| 7 | feature_flag `ENGINE_4_SELF_CORRECTION` = false |
+
+---
+
+## §13 Living Changelog（只追加）
+
+### 2026-04-18 下午 — 本檔建立
+
+**觸發**：MoWoooWorkDown 假警報 RCA 深潛
+**關鍵發現**：
+- 110 load=18.35 / 188 load=16.32 / cadvisor 288% 持續 13 天無告警
+- Prometheus 覆蓋不到三成（35/150+ targets）
+- HostHighCpuLoad 量錯維度（CPU idle vs load_avg）
+- K3s 120/121/125 完全無監控
+
+**統帥決議**：
+- 全景資產盤點 × 7 項自動化 × 永久化 DB
+- AI 四分工（OpenClaw / NemoTron / Hermes / LLM）
+- 拍板執行 Option B（0a+0b 並行止血）+ 全套 Schema Migration（11 張表）
+
+**顧問建議納入**：
+- `asset_inventory.environment` 欄位（blast radius 邊界）
+- `automation_operation_log.retry_count + stderr_feed_back`（引擎 4 閉環）
+
+**下一步**：待統帥授權 Phase 0a/0b 執行
+
+---
+
+## §14 給下一位接手 Claude 的協議
+
+1. **先讀本檔 §0 + §7**（知道現在在哪階段）
+2. **讀 MASTER §0**（AI 自主化北極星）
+3. **讀 CLAUDE.md + HARD_RULES.md**
+4. **跑** `SELECT * FROM asset_discovery_run ORDER BY started_at DESC LIMIT 5`（看最後一次盤點）
+5. **跑** `SELECT host, ai_verdict, recommended_actions FROM host_capacity_snapshot WHERE captured_at > now() - interval '24 hours'`（看 NemoTron 昨夜建議）
+6. **問統帥** 當前 Phase（0a/0b/1/2/3/4/5/6/7）
+7. **絕對不要** 跳過 Phase 順序或在 110/188 load > 10 時跑重審計
+