llm-intelligence/TASKS.md

# 📋 LLM Intelligence Hub Tasks

## 角色
- **产品架构师**：负责需求收敛、P0/P1/P2 划分、文档一致性
- **数据后端**：负责采集器、数据库、日报生成
- **前端实现**：负责 Dashboard / Explorer
- **集成验收**：负责验证脚本、发布条件、日报推送

## T-1 范围收敛
### T-1.1 ✅ Phase 1 范围冻结基线
- **Task**：在 `PRD.md` 中补充 Phase 1 的明确范围、非目标、验收标准
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-09）
- **交付语义**：规划基线完成（不代表后续实现链路验证）
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "Phase 1" /home/long/project/llm-intelligence/PRD.md && grep -q "非目标" /home/long/project/llm-intelligence/PRD.md && grep -q "验收标准" /home/long/project/llm-intelligence/PRD.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

### T-1.2 ✅ 文档冲突清理基线
- **Task**：消除 `PRD.md`、`FEATURE_LIST.md`、`TECHNICAL_DESIGN.md` 中对阶段、技术栈、功能边界的冲突描述
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-09）
- **交付语义**：规划基线完成（不代表后续实现链路验证）
- **verification**:
  - mode: `test_pass`
  - command: `if grep -qE "等待技术设计完成后启动|技术栈待升级" /home/long/project/llm-intelligence/FEATURE_LIST.md /home/long/project/llm-intelligence/TECHNICAL_DESIGN.md; then exit 1; fi; echo clean`
  - expected_evidence: `clean`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

## T-2 数据后端
### T-2.1 ✅ OpenRouter 采集器
- **Task**：新增 `scripts/fetch_openrouter.go`，支持抓取模型基础信息与价格信息
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-08）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go test -tags llm_script scripts/fetch_openrouter.go scripts/fetch_openrouter_test.go -run "TestParseModels|TestRunNoAPIKey" >/tmp/llm_task_t21.out 2>&1 && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 60

### T-2.2 ✅ PostgreSQL migration
- **Task**：新增 `db/migrations`，落地 `models`、`model_prices`、`report_runs` 表
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-06）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase1.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 60

### T-2.3 ✅ 日报生成器
- **Task**：新增日报生成命令，输出 Markdown 报告到 `reports/daily/`
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-07）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase3.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 60

## T-3 前台
### T-3.1 ✅ Explorer 页面脚手架
- **Task**：新增 `frontend/src/pages/Explorer.tsx`
- **Owner**：前端实现
- **状态**：✅ 完成（2026-05-07）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase4.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 90

### T-3.2 ✅ Dashboard 最小组件
- **Task**：提供模型表格、免费标签、价格趋势占位图
- **Owner**：前端实现
- **状态**：✅ 完成（2026-05-07，脚手架就位，组件待完善）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase4.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 90

### T-3.3 ✅ Dashboard 接入套餐订阅价
- **Task**：让 Dashboard 读取 `/api/v1/subscription-plans`，把腾讯云套餐订阅价作为独立区块展示，并与模型价格排行分开展示
- **Owner**：前端实现
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-3.2`、`T-Data-9`
- **结果**：Dashboard 已新增腾讯云套餐订阅价区块、套餐数量摘要和最低月费摘要；前端已支持 `subscription_plan` 归一化、额度格式化和接口异常降级
- **交付语义**：实现完成，代表前端已正式消费套餐订阅价 API；用户无需再只从日报查看腾讯云套餐
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence/frontend && npm test -- --run >/tmp/llm_task_t33_test.log 2>&1 && npm run build >/tmp/llm_task_t33_build.log 2>&1 && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 180

## T-4 OpenClaw 闭环
### T-4.1 ✅ 项目本地任务清单基线
- **Task**：为 `llm-intelligence` 建立独立 `GOALS.md`、`TASKS.md`
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-09）
- **交付语义**：配置基线完成（不代表任务执行链路已验证）
- **verification**:
  - mode: `test_pass`
  - command: `test -f /home/long/project/llm-intelligence/GOALS.md && test -f /home/long/project/llm-intelligence/TASKS.md && grep -q "## G-1" /home/long/project/llm-intelligence/GOALS.md && grep -q "## T-1" /home/long/project/llm-intelligence/TASKS.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `configuration`
  - timeout_seconds: 10

### T-4.2 ✅ 验证器项目本地化
- **Task**：让 `scripts/verification_executor.go` 默认优先读取本项目 `TASKS.md`
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-10）
- **交付语义**：实现完成
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go run -tags llm_script scripts/verification_executor.go --dry-run --tasks /home/long/project/llm-intelligence/TASKS.md | grep -q "/home/long/project/llm-intelligence/TASKS.md" && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 30

### T-4.3 ✅ 项目执行说明基线
- **Task**：沉淀 `OPENCLAW_EXECUTION.md`，说明本项目的角色、协作顺序、验证与回收规则
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-09）
- **交付语义**：规划基线完成（不代表后续实现链路验证）
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "验证真实性协议" /home/long/project/llm-intelligence/OPENCLAW_EXECUTION.md && grep -q "复杂任务执行协议" /home/long/project/llm-intelligence/OPENCLAW_EXECUTION.md && grep -q "Review 产物字段协议" /home/long/project/llm-intelligence/OPENCLAW_EXECUTION.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

## T-5 生产级收口
### T-5.1 ✅ 生产级实施计划基线
- **Task**：将 `IMPLEMENTATION_PLAN.md` 升级为生产级实施计划，显式补充国内厂商覆盖、数据质量规则、容错降级、审计日志
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-10）
- **交付语义**：规划基线完成（不代表各项实现已交付）
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "国内厂商" /home/long/project/llm-intelligence/IMPLEMENTATION_PLAN.md && grep -q "数据质量" /home/long/project/llm-intelligence/IMPLEMENTATION_PLAN.md && grep -q "降级" /home/long/project/llm-intelligence/IMPLEMENTATION_PLAN.md && grep -q "审计日志" /home/long/project/llm-intelligence/IMPLEMENTATION_PLAN.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

### T-5.2 ✅ 任务清单与实施计划基线
- **Task**：补齐 `TASKS.md` 中缺失的生产级收口任务，避免只停留在早期 4 组骨架任务
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-10）
- **交付语义**：规划基线完成（不代表各项实现已交付）
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "生产级收口" /home/long/project/llm-intelligence/TASKS.md && grep -q "环境变量与真实数据链路" /home/long/project/llm-intelligence/TASKS.md && grep -q "前端构建系统初始化" /home/long/project/llm-intelligence/TASKS.md && grep -q "自动采集与日报调度" /home/long/project/llm-intelligence/TASKS.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

### T-5.3 ✅ 环境变量与真实数据链路打通
- **Task**：配置 `OPENROUTER_API_KEY` 与 `DATABASE_URL`，验证真实采集、真实写库、真实日报链路
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-10）
- **结果**：已完成真实 OpenRouter 采集、PostgreSQL 写库和日报生成；`2026-05-10 21:22` 实测 API 拉取 `367` 条，当前库内 `models=377`、`report_runs=2`
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/run_real_pipeline.sh`
  - expected_evidence: `采集完成`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 120

### T-5.4 ✅ 前端构建系统初始化
- **Task**：补齐 `frontend/package.json`、`tsconfig.json`、构建脚本，确保 Explorer 不再只是孤立代码片段
- **Owner**：前端实现
- **状态**：✅ 完成（2026-05-10）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence/frontend && npm run build >/tmp/llm_task_t54_build.log 2>&1 && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 90

### T-5.5 ✅ 自动采集与日报调度
- **Task**：补齐 cron 或等价调度入口，形成真实的每日采集与日报生成闭环
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-10）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase3.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 30

### T-Video-1 ✅ 视频日报原型规划基线
- **Task**：在 `PHASE2_REQUIREMENTS.md` 中冻结视频日报的分类、时长、技术方案和前置依赖
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-11）
- **交付语义**：规划基线完成
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "### 4.1 视频日报（T-Video-1）" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "代码模型日报" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "文本转语音" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "Phase 2 数据源接入" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10
- **后续实现入口**：视频生成 pipeline、TTS 集成和视频验收脚本后续单独拆实施任务

### T-Video-2 ✅ 视频日报生成 pipeline 落地
- **Task**：实现 `scripts/generate_video_digest.go` 或等价入口，按分类生成视频脚本、配音文本和待渲染素材清单
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-11）
- **依赖**：`T-Video-1`、`T-2.3`
- **结果**：已生成 5 组 digest cards、分镜脚本、PNG 帧、manifest、GIF 视频原型和 WAV 旁白音轨
- **交付语义**：实现完成，代表视频日报原型已具备可运行的素材与渲染入口（当前产物为 GIF + WAV 原型）
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go run -tags llm_script scripts/generate_video_digest.go --report /home/long/project/llm-intelligence/reports/daily/daily_report_2026-05-11.md --output-dir /home/long/project/llm-intelligence/reports/daily/video/2026-05-11 | grep -q "cards=5" && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 60

### T-Video-3 ✅ 视频日报端到端验收脚本
- **Task**：补齐视频日报真实验收入口，验证脚本生成、素材产出、音频生成和最终视频文件落盘
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-11）
- **结果**：`verify_video_pipeline.sh` 已能端到端核验 manifest、5 组脚本、5 张 PNG 帧、GIF 视频原型和 WAV 音轨
- **交付语义**：实现完成，代表视频日报原型链路已通过端到端验收
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_video_pipeline.sh`
  - expected_evidence: `VIDEO_PIPELINE: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 120

## Phase 2 数据源扩展需求

### T-Data-1 ✅ 国内云厂商价格采集规划基线
- **Task**：在 `PHASE2_REQUIREMENTS.md` 中冻结国内平台清单、来源区分体系、统一采集接口和 Phase 2 完成条件
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-11）
- **交付语义**：规划基线完成
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "### 3.1 国内模型平台" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "智谱 AI" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "### 3.3 来源区分体系" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "type DataSource interface" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "### Phase 2 完成条件" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10
- **后续实现入口**：各平台采集器、真实 API 验证和 Phase 2 验收脚本后续单独拆实施任务

### T-Data-2 ✅ 多源采集器入口落地
- **Task**：完善 `scripts/fetch_multi_source.go`，让 Moonshot、DeepSeek、OpenAI 等多源采集可独立构建并具备统一入口
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-11）
- **结果**：已支持 `--sources` / `--dry-run` / `--list-sources`，可在不写库的情况下运行静态多源采集并输出摘要
- **交付语义**：实现完成，代表 Phase 2 已具备多源采集执行入口
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go run -tags llm_script scripts/fetch_multi_source.go --dry-run --sources moonshot,deepseek,openai | grep -q "sources=3 successful_sources=3 models=8 domestic_models=5 currencies=CNY:3,USD:5" && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 60

### T-Data-3 ✅ 国内厂商种子与来源字段落库
- **Task**：落地 `source_type/free_quota/free_limitations/rate_limit` 字段和国内厂商种子数据，确保 CNY 定价与来源标识可查询
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-11）
- **依赖**：`T-Data-1`、`T-2.2`
- **交付语义**：实现完成后，才代表 Phase 2 数据模型能承载国内厂商和来源区分
- **verification**:
  - mode: `test_pass`
  - command: `test "$(psql -d llm_intelligence -Atqc "select count(*) from information_schema.columns where table_name='region_pricing' and column_name in ('source_type','free_quota','free_limitations','rate_limit');")" -ge 4 && test "$(psql -d llm_intelligence -Atqc "select count(*) from model_provider where country='CN';")" -ge 7 && test "$(psql -d llm_intelligence -Atqc "select count(*) from region_pricing where currency='CNY';")" -ge 10 && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 60

### T-Data-4 ✅ Phase 2 多源采集验收
- **Task**：以 `scripts/verify_phase2.sh` 为主入口，完成国内厂商覆盖、CNY 定价、采集成功统计和审计记录的真实验收
- **Owner**：集成验收
- **状态**：✅ 完成（2026-05-11）
- **结果**：`verify_phase2.sh` 已通过，国内厂商覆盖、CNY 定价、采集成功统计和模型审计记录均满足 Phase 2 门禁
- **交付语义**：实现完成，代表 Phase 2 数据源扩展已通过集成验收
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && bash scripts/verify_phase2.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 120

### T-Data-5 ✅ 腾讯云公开目录采集入口
- **Task**：新增 `scripts/fetch_tencent_catalog.go` 或等价入口，采集腾讯云公开可见的模型清单、套餐名称、套餐价格、上下文长度、适用范围和来源 URL
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-Data-1`、`T-Data-3`
- **结果**：已新增 `scripts/fetch_tencent_catalog.go`，支持真实 URL 抓取和 `--fixture` 离线 dry-run；当前可解析 `2026-04-27` 公开页快照中的 `8` 个套餐和 `11` 个模型目录项
- **交付语义**：实现完成，代表腾讯云公开目录信息已进入自动采集链路；不代表 Token Plan 套餐已完成现有价格模型映射或正式落库
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go run -tags llm_script scripts/tencent_catalog_lib.go scripts/fetch_tencent_catalog.go --dry-run --fixture /home/long/project/llm-intelligence/scripts/testdata/tencent_token_plan_sample.txt | grep -q "source=tencent-public-catalog updated_at=2026-04-27 17:18:02 plans=8 models=11 series=Hy Token Plan:4,通用 Token Plan:4 dry_run=true" && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `code`
  - timeout_seconds: 60

### T-Data-6 ✅ 腾讯云 Token Plan 套餐映射设计
- **Task**：明确腾讯云 `Token Plan` / `Coding Plan` 的价格模型映射方案，判断是扩展 `region_pricing` 兼容套餐信息，还是新增 `subscription_plan` 表单独承载订阅型价格
- **Owner**：产品架构师
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-Data-5`
- **结果**：已确认 `region_pricing` 继续只承载按模型按量价格，腾讯云 `Token Plan` / `Coding Plan` 单独进入 `subscription_plan`；同时明确了 DDL 草案、映射规则、日报/API 展示边界
- **交付语义**：规划基线完成，代表腾讯云套餐价格的落库路径和验收边界已被明确；不代表数据已自动入库
- **verification**:
  - mode: `test_pass`
  - command: `grep -q "CREATE TABLE subscription_plan" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "为什么不能继续复用" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && grep -q "/api/v1/subscription-plans" /home/long/project/llm-intelligence/PHASE2_REQUIREMENTS.md && echo verified`
  - expected_evidence: `verified`
  - evidence_grade: `artifact-present`
  - task_type: `documentation`
  - timeout_seconds: 10

### T-Data-7 ✅ 腾讯云套餐表迁移与导入
- **Task**：新增 `subscription_plan` 数据库迁移和腾讯云套餐导入入口，把 `fetch_tencent_catalog.go` 解析出的公开目录结果正式落到数据库
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-Data-5`、`T-Data-6`
- **结果**：已新增 `db/migrations/005_subscription_plan.sql` 和 `scripts/import_tencent_subscription.go`；基于公开目录 fixture 已真实落库 `8` 条腾讯云套餐记录
- **交付语义**：实现完成，代表腾讯云公开目录信息已正式落库，可供日报和 API 后续消费
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && grep -q "source=tencent-subscription-import updated_at=2026-04-27 17:18:02 plans=8 provider=Tencent operator=Tencent Cloud table_rows=8 dry_run=false" reports/verification/tencent_subscription_import_latest.txt && go test -tags llm_script scripts/tencent_catalog_lib.go scripts/import_tencent_subscription.go scripts/import_tencent_subscription_test.go >/tmp/llm_tdata7_test.log 2>&1 && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 90

### T-Data-8 ✅ 日报接入腾讯云套餐订阅价
- **Task**：让 `generate_daily_report.go` 读取 `subscription_plan`，在日报 Markdown/HTML 中新增“腾讯云套餐订阅价”区块，并明确该区块不参与按模型价格排行
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-Data-7`
- **结果**：日报生成器已新增腾讯云套餐订阅价区块；`2026-05-13` 的 Markdown/HTML 日报都能展示 `8` 条套餐记录，且未混入模型价格排行
- **交付语义**：实现完成，代表腾讯云订阅型套餐价格已进入日报消费链路，但 API 独立查询入口仍待后续任务
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go test -tags llm_script scripts/generate_daily_report.go scripts/generate_daily_report_test.go >/tmp/llm_tdata8_test.log 2>&1 && TODAY=$(date +%F) && grep -q "## 💳 腾讯云套餐订阅价" reports/daily/daily_report_${TODAY}.md && grep -q "3500万 Tokens/月" reports/daily/daily_report_${TODAY}.md && grep -q "腾讯云套餐订阅价" reports/daily/html/daily_report_${TODAY}.html && echo runtime-ok`
  - expected_evidence: `runtime-ok`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 120

### T-Data-9 ✅ 套餐订阅价独立 API
- **Task**：为 `subscription_plan` 增加独立 API `/api/v1/subscription-plans`，让前端和外部调用方可直接查询套餐数据，而不是只能从日报里读取
- **Owner**：数据后端
- **状态**：✅ 完成（2026-05-13）
- **依赖**：`T-Data-7`
- **结果**：`cmd/server` 已新增 `/api/v1/subscription-plans` 路由与查询逻辑，返回 `subscription_plan` 的 envelope JSON；Phase 6 验收脚本已纳入新接口检查
- **交付语义**：实现完成，代表腾讯云套餐订阅价已具备独立 API 查询入口；前端消费和展示增强仍可后续单独演进
- **verification**:
  - mode: `test_pass`
  - command: `cd /home/long/project/llm-intelligence && go test ./cmd/server >/tmp/llm_tdata9_test.log 2>&1 && bash scripts/verify_phase6.sh`
  - expected_evidence: `PHASE_RESULT: PASS`
  - evidence_grade: `runtime-verified`
  - task_type: `automation`
  - timeout_seconds: 180