| name | qveris-langalpha |
| description | QVeris-native adaptation of candidate 2, LangAlpha. Use for DCF, earnings analysis, earnings preview, and sector overview workflows that preserve LangAlpha-style schemas while routing all financial data through qveris_finance.* CAP tools. |
QVeris LangAlpha
Use this skill for DCF assumptions, sensitivity analysis, earnings post-mortems/previews, and sector overview reports adapted from LangAlpha. Preserve the original workflow categories, but replace fundamentals, market, and macro MCP access with QVeris finance CAP calls.
Source record:
| Field | Value |
|---|
| Candidate number | 2 |
| Original repository | LangAlpha |
| GitHub URL | https://github.com/ginlix-ai/LangAlpha |
| License | Apache-2.0 |
| Evaluation recent activity | 2026-07-06 |
| Local source snapshot | third_party/source_repos/02-langalpha |
| Snapshot latest commit | deab98e on 2026-07-06 |
Runtime Contract
- Use only
qveris_finance.* CAP tools and QVERIS_API_KEY.
- Resolve tickers, exchanges, companies, and CIKs with
ref_symbology, ref_security_master, and ref_company_profile.
- Accept
dry_run, max_calls, max_age, and budget_note; if omitted in a natural-language request, default to dry_run=false, max_calls=12, max_age=P1D, and a conservative budget note, then echo those controls.
- Attach
qveris_trace to every output section and list missing_fields without backfilling.
- Treat QVeris
_meta.source_provider as provenance only; never call, request credentials for, or depend on those internal providers directly.
- Suppress
analyst_target_price, target_price, price-objective, upside, buy/sell, and recommendation fields even if a QVeris payload contains them.
- Sanity-check entity, market, date window, fiscal period, and payload shape before using data; if a payload is stale, cross-period, truncated, or semantically mismatched, mark it in
data_quality and missing_fields.
Workflows
- DCF model:
fundamentals_is, fundamentals_bs, fundamentals_cf, fundamentals_derived_ratios, estimates_consensus, rates_govt_benchmark, mkt_l1_rt.
- Earnings analysis/preview:
event_calendar_earnings, earnings_actual_surprise, estimates_consensus, transcripts_earnings_call, news_fin_realtime.
- Sector overview:
ref_classification_industry, index_constituents, index_levels, flow_sector_capital, mkt_breadth_internals.
Output Requirements
- Use
schemas/output.schema.json for machine-readable output.
- Report assumptions, sensitivity ranges, missing inputs, and confidence.
- Align DCF statement inputs by fiscal year/quarter before calculating; if income statement, balance sheet, cash flow, estimates, or rates arrive on different periods, do not blend them into one scenario table.
- If
rates_govt_benchmark fails, mark the risk-free-rate input missing instead of substituting a stale or non-QVeris value.
- Keep valuation outputs as scenario ranges and assumption audits; do not present target price commitments.
- Include
source_record, controls, analysis, risk_notes, missing_fields, and qveris_trace.
- Include
data_quality with status, stale fields, out-of-window events, and suppressed fields when applicable.
- End with:
不构成投资建议 / Not investment advice.
Prohibited Capabilities
Do not use non-QVeris fundamentals/market/macro MCPs, EODHD, Yahoo, FMP, Alpha Vantage, Polygon, AkShare, Snowball, Sina, SEC scraping, Longbridge, FinViz, Alpaca, browser automation, cookies, login state, third-party API keys, automated trading, wallet/swap, buy/sell points, portfolio action instructions, or target price commitments.
References
- Read
references/qveris-tool-map.md before choosing tool calls.
- Use
fixtures/qveris/sample-output.json as the minimum output shape.
