Claude Skills by tools-only

- [x] 1.1.1 新增 `GET /api/etf/managers` 获取管理人列表 - [x] 1.1.2 新增 `GET /api/etf/tracking-indices` 获取跟踪指数列表 - [x] 1.1.3 新增 `GET /api/etf/trade-dates` 获取可用交易日期 - [x] 1.1.4 扩展 `GET /api/etf/etfs` 支持新筛选参数： - `trade_date`: 指定交易日期 - `manager`: 管理人 - `tracking_index`: 跟踪指数 - `fee_min/fee_max`: 费率区间 - `amount_min`: 最小成交额 - `pct_chg_min/pct_chg_max`: 涨跌幅区间

- [ ] 1.1 Identify all call sites of `sync_task_manager.create_task()` (scheduler + routers) and classify use-cases: incremental, backfill, group execution. - [ ] 1.2 Identify existing status/query endpoints and UI expectations for task status fields (pending/running/completed/failed/cancelled).

- [x] 1.1 创建 `sync_task_history` 表结构（ClickHouse） - [x] 1.2 创建 `src/stock_datasource/modules/datamanage/service.py` 服务层 - [x] 1.3 创建 `src/stock_datasource/modules/datamanage/schemas.py` 数据模型

**Type**: Enumerative **Executor**: LLM **Knowledge**: CWE → CAPEC (Threat Pattern Set) ---

**Type**: Enumerative **Executor**: LLM **Knowledge**: CWE → CAPEC (Threat Pattern Set) ---

**Type**: Enumerative **Executor**: LLM **Knowledge**: CWE → CAPEC (Threat Pattern Set) ---

当前系统有两套互相独立的调度机制，且均未真正生效：

当前系统使用插件化架构管理数据源，每个插件通过 `config.json` 定义调度频率（daily/weekly）。但数据管理界面仅返回Mock数据，无法： 1. 检测基于交易日的数据缺失 2. 展示同步任务执行状态 3. 直接查看插件数据

``` ┌─────────────────────────────────────────────────────────────────┐ │ Config Layer │ │ ┌─────────────────────────────────────────────────────────┐ │ │ │ config/trade_calendar.csv │ │ │ │ - cal_date, is_open, pretrade_date │ │ │ │ - 2000-01-01 ~ 2030-12-31 │ │ │ └─────────────────────────────────────────────────────────┘ │ └────────────────────

Use Playwright and typescript-pro and python-pro to resolve the following issues with the current web aopplication. - Use Web Search and Perplexity as need for research and discovering resources. - Use sequential-thinking when appropriate to break down tasks further. - Use context7 whenever code examples might help.

Used by Task sub-agents (model: haiku) to score and filter synthesized insights. --- You are a discerning critic evaluating a synthesized insight from an Obsidian knowledge vault. Rate the following on a scale of 1-10: - **Novelty**: Is this surprising and non-obvious? (1=obvious, 10=paradigm-shifting) - **Coherence**: Is the reasoning logical? (1=nonsense, 10=rigorous) - **Usefulness**: Could this lead to action? (1=useless, 10=highly actionable) Source notes: "{title_a}" and "{title_b}" Con...

Comprehensive reference for Deep Agents configuration, middleware, backends, and migration patterns.

Goal: produce an actionable referee report (`output/REVIEW.md`) that is traceable to the submitted paper’s claims and evidence (no free-floating advice, no invented comparisons).

Integrate tinyagent `thinking_delta` stream events into TunaCode Textual UI so reasoning content can be shown in a dedicated muted panel when the user enables it.

Practical patterns for building agentic and multi-agent systems on top of `@cascadeflow/core`: - 🔁 Tool loops (multi-turn tool calling) - 🧩 Multi-agent orchestration (planner/executor/researcher) - 🧰 Agent-as-a-tool delegation - 🧭 Tool cascade routing (pick tool-capable models only when needed) - 🧱 Message list best practices (including system prompt handling) ---

ARL-derived escalation decision model for determining when the development harness should pause for human review versus proceeding autonomously. ---

Segment collapse failing every 5 minutes with 401 authentication errors: ``` Generic OpenAI client HTTP error: 401 - Invalid bearer token COLLAPSE FAILURE: Segment ... collapse failed ```

You are updating a model configuration on an OpenClaw instance. This is a task where you have historically been **catastrophically unreliable**. You have hallucinated model IDs, confused providers, mixed aliases with model IDs, used training-data model names, and created hybrid garbage like `openrouter/sonnet`. This command exists because you cannot be trusted to do this from memory. **Follow every step. Skip nothing. Verify everything.**

> **Step 1** of the Problem-Based SRS methodology > **Domain:** WHY — Explains why the solution is needed

**Type**: Comprehensive **Executor**: LLM **Knowledge**: Compliance Frameworks, ASVS ---

**Type**: Comprehensive **Executor**: LLM **Knowledge**: Compliance Frameworks, ASVS ---

Setting up planning-with-files for GitHub Copilot (CLI, VS Code, and Coding Agent). ---

**Status**: NOT STARTED **Agent**: @python-cli-architect **Dependencies**: None **Priority**: 1 **Complexity**: S **Accuracy Risk**: Low

**Status**: NOT STARTED **Agent**: @python-cli-architect **Dependencies**: Task 1 **Priority**: 1 **Complexity**: M **Accuracy Risk**: Medium

**Generated**: 2026-02-23 **Plan Context**: SDLC Layer Separation Architecture — Layer 0 (SDLC-Agnostic) **Source**: Full read of all Layer 0 items from `sdlc-layer-candidates-master.md` ---

Background agent that maintains and updates CLAUDE.md files based on project changes. Invoked at session start and after major milestones (feature completion, refactoring, new dependencies, architecture changes). Works independently without interrupting other agents.

<github_sync>

Reviews workflow-based Claude Code skills for structural quality, pattern adherence, tool assignment correctness, and anti-pattern detection. Use when auditing an existing skill or validating a newly created skill before submission.

How to split skill content across files so the LLM prioritizes correctly. ---

The memory system gives Home Agent persistent long-term memory across conversations. It automatically extracts and stores important facts, preferences, and events, then recalls them when relevant.

How to choose the right tools for skills, agents, and subagents. ---

Updated `system_prompt.md` to align with the current runtime toolset and discovery workflow.

Researches and captures Azure infrastructure project requirements

- **Feature**: github-issue-64 - **Status**: APPROVED - **Created**: 2026-02-04 - **Author**: Factory Design Mode

- **Feature**: launcher-mode-verification - **Status**: APPROVED - **Created**: 2026-01-31

- **Feature**: performance-analysis - **Status**: APPROVED - **Created**: 2026-01-30

- **Feature**: orchestrator-coordination-fix - **Status**: APPROVED - **Created**: 2026-02-04 - **GitHub Issues**: Related to #65, #119 (bite-sized-planning rush failures)

- **Feature**: fix-task-list-id - **Status**: DRAFT - **Created**: 2026-01-31

- **Feature**: pre-release-docs - **Status**: DRAFT - **Created**: 2026-02-01

Step 3 - Design Artifacts. Generates architecture diagrams and Architecture Decision Records (ADRs) for Azure infrastructure. Uses azure-diagrams skill for visual documentation and azure-adr skill for formal decision records. Optional step - users can skip to Implementation Planning.

Expert Azure Bicep Infrastructure as Code planner that creates comprehensive, machine-readable implementation plans. Consults Microsoft documentation, evaluates Azure Verified Modules, and designs complete infrastructure solutions with architecture diagrams.

Expert Azure Bicep Infrastructure as Code specialist that creates near-production-ready Bicep templates following best practices and Azure Verified Modules standards. Validates, tests, and ensures code quality.

Executes Azure deployments using generated Bicep templates. Runs deploy.ps1 scripts, performs what-if analysis, and manages deployment lifecycle. Step 6 of the 7-step agentic workflow.

Generates Step 7 as-built documentation suite after successful deployment. Reads all prior artifacts (Steps 1-6) and deployed resource state to produce comprehensive workload documentation: design document, operations runbook, cost estimate, compliance matrix, backup/DR plan, resource inventory, and documentation index.

Interactive diagnostic agent that guides users through Azure resource health assessment, issue identification, and remediation planning. Uses approval-first execution for safety, analyzes single resources, and saves reports to agent-output/{project}/.

This reference supports clause-by-clause review for commercial NDAs in a jurisdiction-agnostic way. Focus on business risk allocation and operational feasibility.

Define explicit procedure when a task fails irrecoverably. How to undo artifact changes? How to restore artifact plane to consistent state after failure?

MANDATORY template compliance rules for artifact generation

Explore token budgets and cost controls per agent/stage. Track API costs. Set limits per task.

These lessons were learned from analyzing 19+ state privacy laws, comparing automated extraction against professional legal analyses, and conducting cross-jurisdictional compliance projects. ---