A multi-source A-share decision-support skill for screening Shanghai and Shenzhen main-board stocks, analyzing market events, optimizing portfolios, and enforcing trading discipline.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install a-share-expert
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install a-share-expert using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is A-Share Expert?
A-Share Expert is one of the Openclaw Skills built for structured China A-share analysis across nine practical workflows: candidate screening, real-time quotes, fundamentals, capital flows, event-driven research, portfolio optimization, post-market review, theme-launch detection, and rule-based profit-taking or stop-loss management.
The skill coordinates TDX, iFind, mx-ds, FTShare, Wudao, and local Riskfolio tooling according to trading hours and data availability. It validates the latest real-time tick before making current-price or limit-up judgments, filters for the configurable main-board scope, follows deterministic fallback routes, and clearly reports when data cannot be retrieved. It provides decision support rather than trade execution or guaranteed investment advice.
A-Share Expert Use Cases
- Find Shanghai and Shenzhen main-board candidates by market capitalization, price movement, valuation, volume ratio, or theme.
- Check a stock's latest price, percentage change, limit-up status, limit-down status, and pending order value during market hours.
- Review company fundamentals, financial indicators, quarterly results, earnings announcements, and operating data.
- Analyze institutional capital flow, intraday main-force activity, sector strength, and market sentiment.
- Investigate catalysts such as policy events, conferences, capacity changes, commodity prices, and company news.
- Detect newly emerging themes by combining news freshness, sector linkage, volume anomalies, MA20 breakouts, limit-up clustering, and capital-flow confirmation.
- Diagnose an existing position using entry cost, current tick data, predefined profit targets, moving-average rules, time stops, and hard stop-loss levels.
- Generate portfolio rebalancing suggestions with local Riskfolio 7.3 or simplified allocation fallbacks.
- Produce market replay, limit-up review, emotion-cycle analysis, comparison reports, and long-form Markdown documentation.
- Respond to triggers such as stock screening, A-share quotes, holdings, position sizing, theme stocks, stop-loss, take-profit, or disciplined exit planning.
How A-Share Expert Works
- Classify the user's intent into one of nine scenarios, such as screening, single-quote lookup, fundamentals, capital flow, event analysis, portfolio optimization, replay, theme-launch hunting, or rule execution.
- Validate the trading universe before querying data. The default scope is Shanghai and Shenzhen main-board stocks; ChiNext, STAR Market, and Beijing Exchange symbols are excluded unless the TRADING_SCOPE configuration explicitly enables them.
- For every current-price, percentage-change, or limit-up decision, fetch a fresh real-time tick first. TDX is preferred intraday for fast natural-language quote queries, with FTShare or Wudao as fallbacks.
- Inspect the current time window and select permitted sources. Early-session access is restricted mainly to TDX and iFind; active sessions support all configured sources; overnight workflows use historical data, documents, EDB data, and offline Python analysis.
- Query the primary source for the scenario, then retry sequentially with fallback source one and fallback source two. The workflow does not silently fabricate values or issue parallel calls to every provider.
- Validate returned data for freshness, symbol format, market scope, reasonable values, and consistency with the requested scenario.
- Apply the relevant analytical framework: fundamentals and technical context for diagnosis, capital and catalyst evidence for event analysis, multi-factor signals for theme launches, or predefined exit rules for position discipline.
- Return a concise answer, decision checklist, table, or long-form Markdown report. Outputs can include add, reduce, hold, stop-loss, take-profit, watchlist, or position-sizing suggestions, with assumptions and risk limitations stated clearly.
- Cache reusable results for the same stock and time window, and retain context for follow-up questions. If a source remains unavailable, disclose the limitation instead of presenting unverified data.
A-Share Expert Setup
Prerequisites
- An Openclaw runtime capable of loading skills and invoking MCP tools.
- Configured access to the TDX, iFind, mx-ds, FTShare, and Wudao stock-data MCP servers as available in the deployment.
- Local Riskfolio 7.3 with the prepared
/tmp/portfolio_real.pyintegration for portfolio optimization. - Valid six-digit stock codes and, where required, exchange suffixes such as
.XSHEor.XSHG.
Load the skill
Load a-share-expert through the host agent's skill mechanism. After loading, retrieve the stock data source guide for additional provider context:
skill_view('stock-data-source-guide')
For specialized workflows, load the supporting references only when needed:
skill_view('a-share-expert/references/theme-launch-hunting.md')
skill_view('a-share-expert/references/discipline-execution.md')
skill_view('a-share-expert/references/multi-agent-decision-framework.md')
Configure scope and credentials
Set TRADING_SCOPE according to the permitted universe. Keep the default main-board scope unless broader coverage is explicitly required. Ensure each MCP server's credentials and rate limits are configured in the Openclaw environment. Do not substitute AKShare or Tushare for the documented capital-flow paths when those sources are unstable.
Operational checks
- Use
market=AGor an explicit Shanghai/Shenzhen main-board enumeration for TDX calls. - Include
主板or an equivalent exclusion phrase in mx-ds natural-language screening queries. - Pass an explicit
marketsuch asmain_shormain_szto Wudao tools. - Verify iFind token configuration after changes and monitor daily quotas.
- Configure an IM channel and scheduled task only if automated position reminders are required.
A-Share Expert Data Schema & Taxonomy
Core entities
| Entity | Key fields | Typical sources |
|---|---|---|
| Stock quote | code, symbol, price, time, pct_change, volume, is_limit_up, is_limit_down, order_value |
TDX, FTShare, Wudao |
| Stock universe | code, exchange, market_scope, name, sector, concept, float_market_cap |
mx-ds, TDX, Wudao |
| Fundamentals | code, period, revenue, profit, PE, financial_metrics, earnings_status |
iFind, Wudao |
| Capital flow | code, trade_date, main_inflow, main_outflow, net_flow, flow_type |
Wudao, FTShare |
| Catalyst | keyword, event_date, source_time, event_type, news_freshness, price_or_capacity_signal |
Wudao, iFind EDB, news MCP |
| Theme signal | theme, volume_ratio, MA20_breakout, limit_up_count, sector_strength, fund_flow, lifecycle_stage |
Wudao, TDX, iFind |
| Position rule | code, entry_price, cost, current_price, profit_target, moving_stop, hard_stop, time_stop, action |
User input, quote tools, local rules |
| Portfolio allocation | code, returns_history, weights, risk_model, objective |
FTShare, local Riskfolio |
Generated artifacts
- Screening result tables with scope filters, valuation ranges, price movement, volume, and candidate rationale.
- Quote snapshots that begin with a timestamp and the latest price, change, limit status, and order value.
- Fundamental, event-driven, theme, and post-market analysis reports in Markdown.
- Portfolio optimization output as a DataFrame using the
weightscolumn, with 30/15/55 or valuation-based allocation fallbacks when Riskfolio is unavailable. - Position-monitoring summaries comparing live values with four exit levels: +20% partial take-profit, five-day moving-average break, catalyst realization, and -8%/-10% stop-loss thresholds.
- Optional IM reminders and online-document links for reports exceeding approximately 1,024 characters when the host environment supports document output.
Taxonomy
The workflow taxonomy is organized by intent: screening, quote, fundamentals, capital_flow, event_driven, portfolio, replay, theme_launch, and discipline_execution. Every result should also retain market_scope, data_source, retrieval_time, fallback_level, and confidence_or_limitation metadata.
A-Share Expert Advanced Features
- Time-aware MCP routing that selects sources based on early-session, intraday, post-market, or overnight availability.
- Mandatory real-time tick validation before current-price, limit-up, or profit-taking decisions.
- Deterministic primary-to-fallback source chains with explicit failure reporting and no invented data.
- Configurable trading scope enforcement across TDX, mx-ds, Wudao, and related tools.
- Theme-launch detection using news freshness windows, catalyst calendars, sector linkage, volume-ratio anomalies, MA20 breakouts, limit-up clustering, capital flow, and theme lifecycle stages.
- Rule-based discipline execution with partial take-profit, moving stop, catalyst realization, hard stop-loss, and time-stop conditions.
- Scheduled position checks, such as a 14:30 daily review or 30-minute intraday scans, with optional IM notifications.
- Local Riskfolio 7.3 integration using
port.optimization(model='Classic', rm='MV', obj='MinRisk')and aweightsDataFrame result. - Multi-agent decision support through anti-hallucination checks, role opposition, five-level ratings, and decision retrospectives.
- Result caching for repeated stock queries within the same time window.
- Structured Markdown tables for prices, changes, portfolio weights, evidence, triggers, actions, position sizes, and stop levels.
- Specialized coordination with attribution, auction-review, theme-research, and stock-data-guide skills through conditional
skill_viewloading. - Long-report handling that separates concise chat responses from optional online documents to prevent IM message flooding.
SKILL.md
Loading
Related Openclaw Skills
Huawei Cloud CCE Cluster List
A read-only Openclaw Skills integration for inventorying Huawei Cloud CCE clusters by name, ID, status, version, and flavor.
Post-Fire Vegetation Recovery
A local geospatial analysis skill that maps five-level burn severity with dNBR and tracks vegetation recovery from post-fire NDVI time series.
Amazon Listing Evidence Audit
Amazon Listing Evidence Audit evaluates listing readiness and turns unsupported claims, unclear content, keyword issues, and missing visual proof into a prioritized action plan.
Resume Styles Kit
Resume Styles Kit transforms a candidate’s real experience into complete, polished, and customizable PDF resumes without inventing or removing meaningful details.
Playwright Stealth Verify
Playwright Stealth Verify measures whether an automated browser presents a coherent, internally consistent fingerprint without modifying or evading detection.
DataQuant Connector
DataQuant Connector gives AI agents a secure REST API workflow for retrieving multi-market financial data, screening securities, and supporting quantitative research.
Featured*
