Open QVeris Skills

This repository hosts QVeris (qveris.ai) related bot/agent skills.

• Feel free to use these skills: Use them as-is, fork them, and adapt them for your own bots/agents.
• Contributions welcome: New skills, improvements, fixes, and documentation updates are encouraged.

Repository structure

• One skill per folder: Each top-level folder represents a standalone skill.
• qveris-official/: The official QVeris skill (core capabilities — semantic tool discovery & unified execution).
  • Use it as a starting point: you can modify it (or ask your AI to modify it) to create novel, useful skills tailored to your workflow.
• QVeris A-share beta skills:
  • qveris-a-stock-data-layer/: Broad China A-share data-layer reports for quotes, bars, fundamentals, news, events, flow-like context, and specialty A-share fields. Specialty fields degrade honestly when QVeris CAP evidence is missing.
  • qveris-a-share-factor-screen/: China A-share research screens and factor coverage notes. It produces research candidate pools, not rankings or recommendations when evidence is not comparable.
  • qveris-a-share-data/: China A-share quote, history, technical-context, event, sector, A+H/IPO timeline, and news data reads. Technical indicators are descriptive and never trade signals.
• QVeris third-party finance beta skills:
  • qveris-alphaear-market-intelligence/: QVeris-only adaptation of #42 Awesome Finance Skills / AlphaEar for stock context, news, sentiment availability, signal-monitor notes, and Markdown reports.
  • qveris-daymade-financial-data-suite/: QVeris-only adaptation of #37 Daymade Financial Suite for financial data packs, news/research/event structuring, A-share news reads, and sector or pharma daily monitors.
  • qveris-uzi-equity-research/: QVeris-only adaptation of #34 UZI-Skill for equity research method audits, valuation-input coverage, LHB/flow context, and trap-risk monitoring without investment advice.
  • qveris-crypto-market-radar/: Read-only adaptation of #23 GMGN Skills for crypto identity, spot, history, descriptive technical context, rankings, market mood, and whale monitoring. Wallet, swap, order, and signing features are removed.
• stock-copilot-pro/: A standalone global stock analysis skill for ClawHub/OpenClaw style agents.
  • Includes multi-source routing (quote, fundamentals, technicals, sentiment), quality checks, and structured reports.
• third_party/: Evaluation area for third-party skills being adapted into QVeris Featured Skills.
  • Candidates must have a permissive license, visible quality/traction, and a clear QVeris data-acquisition path.
  • Discovery and tracking are handled by qcli skill third-party from the quaestio-cli repository.
• Planned skills: Development and testing are currently underway.
  • exchange-rate/ – Real-time forex rate and amount conversion.
  • stock-analyzer/ – A-share/global quote, fundamentals, history, smart stock picking.
  • crypto-tracker/ – Crypto price, markets, conversion (CoinMarketCap, CoinGecko).
  • news-briefing/ – News and WeChat article search (caidazi, xiaosu, X news).
  • location-service/ – POI and IP location (Amap, Baidu Map).
  • movie-guide/ – Movie/TV details and reviews (IMDb, NYT).
  • developer-tools/ – IP lookup and geo.
  • content-discovery/ – Taobao, Xiaohongshu, Kuaishou search.
  • weather-alerts/ – Weather alerts and air quality (AQI) by location.
  • term-translation/ – Term/phrase translation (e.g. Data.gov).
  • food-nutrition/ – Food search and nutrition (USDA FoodData Central).
  • sports-score/ – Sports leagues, games, standings (API-Sports, Api-Football).

Getting started

• Pick a skill folder (for example qveris-official/) and follow its README:
  • See qveris-official/README.md
• Skills that call QVeris APIs typically require:
  • QVERIS_API_KEY (get one from qveris.ai)

Using the A-share beta skills

Install qveris-official plus the desired A-share skill folders into your Codex skills directory, or run Codex from an environment that indexes this repository’s skill folders:

mkdir -p ~/.codex/skills
cp -a qveris-official ~/.codex/skills/
cp -a qveris-a-share-data ~/.codex/skills/
cp -a qveris-a-share-factor-screen ~/.codex/skills/
cp -a qveris-a-stock-data-layer ~/.codex/skills/
cp -a qveris-alphaear-market-intelligence ~/.codex/skills/
cp -a qveris-daymade-financial-data-suite ~/.codex/skills/
cp -a qveris-uzi-equity-research ~/.codex/skills/
cp -a qveris-crypto-market-radar ~/.codex/skills/

Standalone installs should either run in a Codex runtime that exposes native qveris_finance.* tools or include qveris-official. The repository CLI fallback (qveris-official/scripts/qveris_tool.mjs) is a repo-root development path; a copied QVeris finance skill folder alone is not enough for that fallback.

Example prompts:

Use qveris-a-share-data to make a 30-day quote, bars, technical-context, event, and news read for 600519.SH.
Use qveris-a-share-factor-screen to run a research screen for 600519.SH, 000001.SZ, and 000858.SZ.
Use qveris-a-stock-data-layer to make an A-share data-layer coverage note for 600519.SH.
Use qveris-alphaear-market-intelligence to make a TSLA market-intelligence note with sentiment availability and trace appendix.
Use qveris-daymade-financial-data-suite to collect an NVDA QVeris-only financial data pack with missing_fields and data_quality.
Use qveris-uzi-equity-research to audit valuation-method input coverage for 600519.SH without target output or trading action.
Use qveris-crypto-market-radar to compare BTC, ETH, and SOL over one aligned 30-day window without transaction actions or forecasts.

Current beta caveat: A-share CAP coverage is partial. Live verification on 2026-07-08 showed usable evidence for quotes, EOD bars, industry classification, calendar events, lock-up, stock-level order-size flow, top movers, IPO calendar context, share structure, and tagged news. Some routes still degraded or failed in smoke tests, including LHB, northbound/cross-border flow, concept heat, CN ETF option chain, CAP-based sentiment score, CAP-based technical indicators, non-empty research reports, and non-empty news clusters. The skills are designed to mark these fields missing instead of inventing data.

Current third-party finance beta caveat: #42, #37, and #34 preserve workflow intent from the source repositories but do not carry their direct data dependencies into runtime. AlphaEar forecasts, Daymade specialty pipelines, and UZI LHB/trap layers are QVeris-only and conditional; when a QVeris CAP is unavailable, stale, empty, or semantically mismatched, the skills mark missing fields instead of inventing a full analysis.

Local development (not part of published skill package)

Local integration files live under dev-infra/ and are intentionally kept outside skill folders such as stock-copilot-pro/.

Reuse repository root .env.local for all local workflows (skill-dev and optional openclaw compose profile). No extra .env files are required under dev-infra/.

• Bring up dev container: make up
• Check runtime: make check
• Run smoke regression (allow-degraded for live API variance): make smoke
• Bring up OpenClaw too: make up-full
• Stop containers: make down

Repository root .env.local is git-ignored by default to avoid accidental key commits.

Contributing

• Add a new folder for a new skill (or improve an existing one).
• Include clear documentation (a README.md and/or SKILL.md) describing:
  • What the skill does
  • Required environment variables / credentials
  • Example prompts / usage

Third-party Skill Discovery

Use qcli skill third-party from the quaestio-cli repository to search GitHub for skill candidates, score them, save the analysis to the admin database schema, and scaffold evaluation branches:

uv run qcli skill third-party apply-db --db-url "$ADMIN_DATABASE_URL"

GITHUB_TOKEN=... \
uv run qcli skill third-party discover \
  --min-stars 50 \
  --per-query 10 \
  --write-db \
  --db-url "$ADMIN_DATABASE_URL" \
  --notify

Candidates are evaluated under third_party///. A candidate can become a Featured Skill only after direct third-party data access has been replaced by QVeris tool search/execute calls and real smoke tests pass.

License

MIT — see LICENSE.