qveris-a-share-data

QVeris A-Share Data

Use this skill to preserve the A-Share Skill candidate's A-share research-data workflows while replacing local market packages, source-specific scripts, and short-term trading modules with QVeris structured-data evidence, a narrow audited Web Search lane for news and qualitative sentiment, and explicit data-quality controls.

Source record:

Source Adaptation

• Preserve the original a-share-data workflows: real-time quote, historical bars, technical indicators, corporate events, A+H list, A-to-HK IPO timeline, hot industry/concept reads, market news, and sector info.
• Replace original Python scripts and dependencies (akshare, MyTT, pandas, numpy, requests) with QVeris CAP calls plus calculated indicators from validated QVeris bars. Use Web Search only for the temporary news and qualitative-sentiment lane defined below.
• Treat original source fallbacks such as direct market HTTP endpoints, sector APIs, DangInvest, and source-specific caches as migration context only.
• Remove the repository's trading modules from this QVeris skill: short-line trading, MACD trading plans, paper trading, accounts, orders, backtests, position discipline, stop-loss, and entry/exit rules.
• Keep technical indicators descriptive. Do not turn MACD, RSI, MA, or BOLL into a buy/sell/position signal.

Runtime Contract

• Use only qveris_finance.* CAP tools and QVERIS_API_KEY for identity, quotes, bars, classifications, and events. Web Search is allowed only for issuer news and qualitative sentiment under the Web News And Sentiment Lane.
• Execute every finance data call through this Skill's scripts/qveris_finance_adapter.mjs, or through a native wrapper that runs the byte-identical adapter; never call /capabilities/query directly from the workflow.
• Default natural-language output to a Markdown user report, not a JSON object.
• Accept dry_run, max_calls, max_age, budget_note, and source_mode; if omitted, default to dry_run=false, no hard max_calls limit, max_age=P1D, a conservative budget note, and source_mode=hybrid_web_news_sentiment, then echo those controls. Benchmark and replay runs use the same narrow Web lane for news and qualitative sentiment; replay reads only frozen Web evidence.
• Read and follow references/qveris-web-news-sentiment-policy.md. It supersedes any older news/sentiment fallback text in this Skill.
• Read references/qveris-finance-data-quality-rubric.md before using QVeris payloads as evidence.
• Read references/qveris-workflow-semantic-guards.md before any derived, comparative, ranking, sentiment, or multi-layer workflow. Use scripts/qveris_workflow_guards.mjs for the applicable pre-prose gates.
• Use references/qveris-finance-retry-policy.md for failed calls, invalid capabilities, payload truncation, and semantic mismatches.
• Build trace, call counts, retries, and timestamps only from saved observed_calls. Never invent an execution ID or planned call; use execution_id=null when an observed call returned no ID.
• Record every Web Search and opened page used for news or sentiment with query, final URL, publisher, publication time, access time, issuer-match result, window-match result, and body-content SHA-256. Search-result snippets alone are not evidence.
• Sanitize every output surface, including Evidence, Sources, prose, params, responses, and Trace. Strip provider names, provider API URLs, raw route/tool IDs, candidates, failover, credentials, and routing metadata recursively; the Trace row remains exactly tool_name, params, status, execution_id, fallback_used, and missing_fields.
• Strip the original candidate's short-term trading and paper-trading behavior. This skill only supports research data reads.
• Suppress target prices, upside/downside, ratings, buy/sell wording, rebalancing instructions, and trade execution plans even if present in QVeris payloads.