Files
uncle6me-web 6a75117ba3 feat(kbdb): recipe 公庫/私庫雙向機制 + UUID 身份 + KBDB Base + 市場數據
kbdb-base SDD §7.5(公庫/私庫雙向機制,richblack 2026-06-07 拍板)。

## KBDB Base worker(新)
- kbdb/:D1-only 核心三表(entries/templates/entry_values)+ CRUD + LIKE search
  + recipe-stats 端點(市場數據)+ 0001_base.sql migration(含 recipe_stat seed)

## Phase 2.3:init 建 D1 + 套 migration
- cli cf-api.ts 加 listD1Databases/ensureD1Database;init 建 arcrun-kbdb D1
- deploy.ts 部署後對 D1 套 0001_base.sql(CF /d1/query API,idempotent)+ 注入 database_id

## Phase 5.1:recipe 成功記錄(市場數據來源)
- GraphExecutor 收集本次用到的 recipe uuid(usedRecipeKeys)
- executeWebhookGraph 執行結束一次性記 per-uuid 成功/失敗到 KBDB(fire-and-forget)

## Phase 7.5:recipe UUID 身份 + app-store 模型
- recipe 領 uuid=唯一身份;canonical_id/author/公私=屬性(§7.5.5)
- recipe:{uuid} + idx:canonical/installed/hash;resolveRecipe 向後相容不破執行鏈
- POST /recipes/submit=領新 uuid 新增作者版本(非覆蓋,app-store)
- GET /public-recipes 搜尋(多作者+per-uuid 市場星數)/ :id pull(選市場最佳)
- 落空→found:false 創作引導(§7.5.6 閉環)
- POST /recipes/migrate-uuid 一次性轉舊 key(增量寫不刪舊、冪等)
- init-seed 用 UUID(author=system)

## 薄殼(rule 07 §5:CLI + MCP 覆蓋同組能力)
- CLI: acr recipe search/pull/submit-p(config 加 DEFAULT_PUBLIC_LIBRARY_URL)
- MCP: arcrun_recipe_search/pull/submit_p/push/list/delete(補齊漂移)

## 壓測修正
- api-recipe-seeds: google_sheets_append PUT→POST(:append 正確動詞,階段12)

四 worker tsc 全綠(cypher/cli/kbdb/mcp)。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-07 16:18:10 +08:00

89 lines
3.7 KiB
SQL

-- KBDB Base — atomic universal table (3 tables, never changes)
-- SDD: .agents/specs/arcrun/kbdb-base/design.md
--
-- Plugin model (like PostgreSQL core + PGVector/AGE):
-- - Base = these 3 tables + plain CRUD + D1 LIKE search (D1 only, free, no credit card).
-- - embed module = optional, writes vectors to Vectorize (does NOT alter these tables).
-- - triplet module = separate repo, writes derived records into entry_values (does NOT alter base).
-- "Table never changes": new tech records its output elsewhere, never ALTERs the base.
-- ============================================================
-- 1. Three tables
-- ============================================================
-- Universal main table: each entry is one atomic datum.
-- entry_type extended for arcrun: 'block' | 'value' | 'template' | 'slot' | 'project' | 'workflow' | 'recipe_stat'
CREATE TABLE IF NOT EXISTS entries (
id TEXT PRIMARY KEY,
content TEXT,
entry_type TEXT NOT NULL,
owner_id TEXT, -- multi-tenant: namespace (self-hosted) or api_key (SaaS)
-- tree structure (project -> workflow via parent_id; SDD Q1 decision)
parent_id TEXT,
-- optional block metadata (harmless plain columns)
page_name TEXT,
refs_json TEXT DEFAULT '[]',
tags_json TEXT DEFAULT '[]',
task_status TEXT,
-- optional embed bookkeeping (set by optional embed module; base never reads them)
content_hash TEXT,
is_embedded INTEGER DEFAULT 0,
-- metadata
confidence REAL,
metadata_json TEXT,
created_at INTEGER DEFAULT (unixepoch()),
updated_at INTEGER DEFAULT (unixepoch())
);
-- Template table: defines slots for a virtual table
CREATE TABLE IF NOT EXISTS templates (
id TEXT PRIMARY KEY,
name TEXT UNIQUE NOT NULL,
description TEXT,
slots_json TEXT NOT NULL, -- JSON array, e.g. ["display_name","gender"]
created_by TEXT, -- 'system' | 'ai' | owner_id
created_at INTEGER DEFAULT (unixepoch()),
updated_at INTEGER DEFAULT (unixepoch())
);
-- Slot link table: composes multiple entries into one structured record
CREATE TABLE IF NOT EXISTS entry_values (
id TEXT PRIMARY KEY,
record_id TEXT NOT NULL,
template_id TEXT NOT NULL REFERENCES templates(id),
slot_name TEXT NOT NULL,
entry_id TEXT NOT NULL REFERENCES entries(id),
created_at INTEGER DEFAULT (unixepoch()),
UNIQUE(record_id, slot_name)
);
-- ============================================================
-- 2. Indexes
-- ============================================================
CREATE INDEX IF NOT EXISTS idx_entries_type ON entries(entry_type);
CREATE INDEX IF NOT EXISTS idx_entries_owner ON entries(owner_id);
CREATE INDEX IF NOT EXISTS idx_entries_parent ON entries(parent_id);
CREATE INDEX IF NOT EXISTS idx_entries_page ON entries(page_name);
CREATE INDEX IF NOT EXISTS idx_entries_task ON entries(task_status);
CREATE INDEX IF NOT EXISTS idx_entries_hash ON entries(content_hash);
CREATE INDEX IF NOT EXISTS idx_ev_record ON entry_values(record_id);
CREATE INDEX IF NOT EXISTS idx_ev_template ON entry_values(template_id);
CREATE INDEX IF NOT EXISTS idx_ev_template_slot ON entry_values(template_id, slot_name);
CREATE INDEX IF NOT EXISTS idx_ev_entry ON entry_values(entry_id);
-- ============================================================
-- 3. Seed templates used by arcrun base
-- ============================================================
-- recipe_stat: success/failure counters for a recipe (feeds recipe submission, SDD section 7)
INSERT OR IGNORE INTO templates (id, name, description, slots_json, created_by)
VALUES
('tpl-recipe-stat', 'recipe_stat', 'recipe success/failure counters', '["canonical_id","success_count","failure_count","last_status","last_at"]', 'system');