Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →asgard-ai-platform/med-business
Use when writing financial news — earnings, IPO/M&A, regulatory changes, industry trends — from transcripts, filings, or investor materials. Activates business-beat workflow: financial-figure provenance, analyst-stake disclosure, securities-law risk, and number stewardship. Triggers on '寫一篇財報季報導', '分析這份財報', '整理 IPO 新聞', 'turn earnings call into a piece', 'write up M&A announcement'. Defers general craft to med-news-reporter.
Passed 19 of 22 test cases with the skill, and 6 of 22 without it.
and 16 more in the full report
13 fixed6 passed either way3 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
Semiconductor Foundry Co reports a Price-to-Book ratio of 4.5x. Draft a business news report paragraph on its financial valuation.
| Checked | Without skill | With skill |
|---|---|---|
| Avoids asserting valuation judgment on 4.5x Price-to-Book without comparison context. | Missed | Passed |
| Includes valuation assertion checks in the mandatory metadata audit footer. | Missed | Passed |
The instructions your model reads when the skill loads.
This skill specializes med-news-reporter for the business beat. Read med-news-reporter first for the general 6-step workflow (type selection, material audit, fact-check, balance, ethics, literacy). This file adds business-beat-specific discipline on top.
Distilled from financial-journalism curricula at NCCU 新聞系、NTU 新聞研究所、Columbia Journalism Review (Business Desk), Medill (Business Journalism), CUNY Newmark, Reuters Institute. Covers the four main business-news sub-types: earnings / IPO-M&A / regulatory-policy / industry-trend reporting. Core mandate: every financial figure must trace to base period, scope (consolidated vs standalone; domestic vs group), source (CFO statement / 公開資訊觀測站 / analyst estimate), and confidence level (audited vs preliminary vs forecast). The non-obvious failure is when the literal number is correct but the missing context lets readers compare apples-to-oranges.
IRON LAW: Numbers Need Stewardship
Every financial figure in the piece — revenue, EPS, growth rate, margin,
valuation multiple — must explicitly carry FOUR metadata tags: (a) BASE
PERIOD / FISCAL YEAR (e.g., "113年度合併報表"), (b) SCOPE (consolidated
vs individual; domestic vs group; operating vs including-financial gains),
(c) SOURCE (CFO statement / 公開資訊觀測站 / analyst consensus / research
report), and (d) CONFIDENCE (audited / preliminary / forward-looking).
Without all four, the figure is marked [待查證] or omitted. Default LLM
behavior is to drop context for "cleaner prose" or assume implied scope
— this suppresses that default by requiring explicit stewardship. The
reader cannot infer scope from a number alone; ambiguity is not efficiency.
Why this is non-obvious: a headline "Company posts record 5B EPS" is technically true (audited, consolidated, 2025 FY) but sounds miraculous without the context that it's 50% higher than last year due to a one-time merger gain, not operating performance. When readers hear "record EPS", they assume baseline comparable operating earnings — a fundamental mismatch. Stewardship audit catches this by forcing every number to carry its scope flags.
Rationalization Table — these justifications DO NOT override the Iron Law:
| Claude might think... | Why it's still a violation |
|---|---|
| "EPS growth was 30%, that's the key number, we don't need to explain consolidated vs individual" | Scope ambiguity is the #1 source of misleading financial journalism. The reported-to-media EPS (often consolidated, including one-time items) differs from operating EPS. Both numbers are "true"; omitting the qualifier lets readers confuse them. |
| "The press release says 營收成長 15%, I'll use that without checking if it's year-over-year or 2-year stack" | Base-period ambiguity is a gotcha. Always verify: Is 15% versus prior-year Q2 or full year? Is the comparison to 113年同期 or 累計? Mismatched bases are the #2 error. |
| "Analyst estimates are just opinions, not facts, so I can soften them as 'expected' instead of marking them as forecasts" | Analyst consensus is market material — forecast signals must be clearly marked as such. When you write 'expected to grow', readers hear 'will grow'. Use 'X firm projects' / 'consensus estimate' to signal uncertainty. |
| "The company's 本益比 of 12 is reasonable, so I can describe it as 'fairly valued' " | 本益比 is a metric, not a judgment. "Fairly valued" requires comparison to peer multiple, historical average, or growth rate context. The bare number 12× is incomplete. Omit judgment or add comparison. |
| "I'll round the margin from 24.3% to 'about 24%' for readability" | Rounding financial figures obscures precision. Use the reported number or mark as [estimated]. Readers trust journalists with numbers; small rounding is still fabrication. |
| "Market cap of 2.5T NT$ is big, readers don't need to know the share price and shares outstanding" | Implied scope: are we talking market cap at close? Intra-day? Converted at what exchange rate if international? Stating the date and share price (+count) lets readers verify and understand concentration risk. |
test cases passed, vs 6 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.