@v587d/capital-generation 2.10.2 → 2.11.0

package/selected-skills/peter-lynch-skill/skills/peter-lynch-skill/references/deep/web-research-1.md

@@ -56,7 +56,7 @@ value, currency/units, source ID(s), and label: `Reported`, `Calculated`, or
 
 Final report citations must use Source Ledger URLs and this format:
 
-`source: [Source name](https://...)`
+`source: [Source name as returned](link returned by retrieval)`
 
 ## Budget
 

package/selected-skills/peter-lynch-skill/skills/peter-lynch-skill/references/deep/web-research-2.md

@@ -8,13 +8,10 @@ Use annual-report discovery targets first — each line below is one thing to as
 the host's retrieval channel for, worked through in order and stopped once a
 complete, current annual report or equivalent full-year filing is in the ledger:
 
-* `"{company name}" investor relations annual report site:{company-domain}`
-* `"{company name}" annual report {current year OR prior year} investor relations`
-* `"{company name}" 10-K site:sec.gov`
-* `"{company name}" 20-F site:sec.gov`
-* `"{ticker}" SEC 10-K 20-F annual report`
-* `"{company name}" annual report PDF`
-* `"{company name}" integrated report`
+* the issuer's own annual report, current or prior fiscal year
+* the annual report filed with the regulator (10-K, or 20-F for a foreign private issuer)
+* the integrated report or equivalent full-year disclosure, when the issuer publishes one
+* the prior year's annual report, for period-over-period comparison
 
 Selection rules:
 

package/selected-skills/peter-lynch-skill/skills/peter-lynch-skill/references/quick/system-prompt.md

@@ -31,7 +31,7 @@ Your report MUST be clean and consistent for **GitHub-Flavored Markdown (GFM)**
 
 - Every company description, financial number, valuation metric, market data point, management claim, industry claim, competitor claim, ownership claim, or news/update claim must include a linked source.
 - Place source lines directly below the bullet or paragraph they support.
-- Use this exact format: `source: [Source name](https://...)`
+- Use this exact format: `source: [Source name as returned](link returned by retrieval)`
 - If one bullet uses multiple sources, list them on the same source line separated by semicolons.
 - Do not cite sources inline at the end of sentences.
 - Do not create a detached source list as a substitute for claim-level source lines.

package/selected-skills/peter-lynch-skill/skills/peter-lynch-skill/references/quick/web-research-1.md

@@ -18,97 +18,98 @@ blocks the screen.
 
 ## Minimum Source Rules
 
-Use reputable aggregator sources for Quick Research. Preferred source types:
-
-1. financial-data aggregators with multi-year statements and key ratios, such
-   as Yahoo Finance, MarketWatch, Macrotrends, StockAnalysis, GuruFocus, or
-   similar public financial-data pages
-2. market-data aggregators with quote, market capitalization, P/E, dividend,
-   and 52-week range data, such as CompaniesMarketCap, Nasdaq, Yahoo Finance,
-   MarketWatch, StockAnalysis, GuruFocus, or similar pages
-3. aggregator or quote/profile pages that include a short business description,
-   sector, industry, exchange, and country
-
-Do not use annual report PDFs, full 10-K / 10-Q / 20-F extraction, SEC filing
+Use public summary sources for Quick Research. Preferred evidence types:
+
+1. financial-data pages with multi-year statements and key ratios, where the
+   page states each period and the reporting currency
+2. market-data pages with a current quote, market capitalization, P/E,
+   dividend, and 52-week range, timestamped
+3. profile or quote pages that include a short business description, sector,
+   industry, exchange, and country
+
+Do not use annual report PDFs, full 10-K / 10-Q / 20-F extraction, filing
 detail pages, company quarterly reports, or long investor-relations documents
 for Quick Research. Those belong to Deep Research.
 
 Never use anonymous blogs, forums, social media posts, or unsourced pages for
 financial numbers. Do not use news articles as the primary source for standard
-financial metrics when an aggregator table is available.
+financial metrics when a structured summary table is available.
 
 Every retrieval request must fill a Source Ledger slot, resolve a
 decision-critical conflict, or support a triggered check. Do not ask for another
 source out of curiosity.
 
-Do not ask source-by-source per fiscal year or per metric when one
-aggregator table can answer the question. Extract revenue, earnings per share
-(EPS), margins, cash, debt, shares, dividends, buybacks, and cash-flow lines
-from aggregator pages.
+Do not ask source-by-source per fiscal year or per metric when one summary table
+can answer the question. Extract revenue, earnings per share (EPS), margins,
+cash, debt, shares, dividends, buybacks, and cash-flow lines from the same
+returned page set.
 
 Before asking for a fourth source after the third Quick retrieval, ask internally:
 
 * What decision-critical fact is missing?
-* Can another section of an existing aggregator source answer it?
-* Which reputable aggregator is most likely to contain the missing metric?
+* Can another section of an existing source answer it?
+* Which evidence type above most plausibly contains the missing metric?
 
-Ask for more only if another aggregator is needed to complete the existing Quick
-methodology responsibly.
+Ask for more only if another summary source is needed to complete the existing
+Quick methodology responsibly.
 
-If aggregator sources conflict, prefer the source that is more recent, more
+If returned sources conflict, prefer the one that is more recent, more
 complete, and more directly tied to the metric. Note material conflicts in the
 brief. Do not resolve normal Quick conflicts by opening full filings; if the
-conflict is decision-critical, ask for another reputable aggregator or move the
+conflict is decision-critical, ask for another summary source or move the
 decision to WATCHLIST / DEEP RESEARCH with the conflict stated.
 
 Every important number in the research brief must include metric, period/date,
 value, currency/units, source ID(s), and label: `Reported`, `Calculated`, or
 `Inference`.
 
-Final report citations must use Source Ledger URLs and this format:
+Final report citations must use Source Ledger entries and this format:
 
-`source: [Source name](https://...)`
+`source: [Source name as returned](link returned by retrieval)`
 
 ## Budget
 
 Practical default:
 
-* **1-3 aggregator sources**
+* **1-3 summary sources**
 * **2-3 retrieval requests**
-* **2-3 page fetches**
+* **2-3 page reads**
 
-Expand only when another reputable aggregator is needed for missing required
-Quick evidence, ambiguous identity, live valuation, a possible fatal flaw, a
-triggered special check, or a decision-critical conflict.
+Expand only when another source is needed for missing required Quick evidence,
+ambiguous identity, live valuation, a possible fatal flaw, a triggered special
+check, or a decision-critical conflict.
 
 ## Default Source Set
 
-Use the smallest aggregator source set that completes the screen:
+Use the smallest returned source set that completes the screen:
 
-1. One aggregator with financial statements or financial history.
-2. One aggregator with live valuation and market data.
-3. One aggregator profile/summary source only if the first two do not cover
-   identity, business description, sector, industry, or exchange.
+1. One source with financial statements or financial history.
+2. One source with live valuation and market data.
+3. One profile/summary source only if the first two do not cover identity,
+   business description, sector, industry, or exchange.
 
-Ownership, insider, competitor, or industry aggregator sources are allowed only
-when the existing Quick methodology needs that evidence and it could affect
-conviction.
+Ownership, insider, competitor, or industry sources are allowed only when the
+existing Quick methodology needs that evidence and it could affect conviction.
 
 ## Query Pack
 
-Each line below is one thing to ask the host's retrieval channel for, not a page
-to go and open from here. Work through them in order and stop as soon as the
-Source Ledger covers required Quick evidence:
+Each line below is one evidence item to request from the host's retrieval
+channel, not a page to open from here. Work through them in order and stop as
+soon as the Source Ledger covers required Quick evidence:
 
-* `"{ticker}" Yahoo Finance financials statistics profile`
-* `"{ticker}" StockAnalysis financials balance sheet cash flow ratios`
-* `"{ticker}" MarketWatch financials income balance sheet cash flow`
-* `"{ticker}" Macrotrends revenue EPS cash flow debt`
-* `"{ticker}" CompaniesMarketCap market cap PE ratio dividend 52 week range`
-* `"{ticker}" Nasdaq quote financials institutional insider`
+* multi-year income statement, balance sheet, cash-flow and ratio history for
+  the ticker
+* current quote, market capitalization, P/E, dividend yield and 52-week range
+* business description, sector, industry, exchange and country
+* ownership and insider activity, only when a triggered check needs it
+* competitor or industry comparison, only when a triggered check needs it
 
 If the company has a non-US listing, add exchange, country, or reporting
-currency. If the ticker is ambiguous, ask for ticker plus exchange name and use
-aggregator profile pages to resolve the traded entity.
+currency. If the ticker is ambiguous, ask for ticker plus exchange name and
+resolve the traded entity from the returned profile.
+
+Ask for the return fields with the request: the figure, the period it covers,
+the reporting currency, the retrieval path, and the source name and link. A
+number that comes back without them does not enter the Source Ledger.
 
 Do not run the full pack automatically.

package/selected-skills/quant-agent-skills/UPSTREAM.md

@@ -109,3 +109,11 @@
   **从不执行、也从不被 provider 读取**（skills 根里没有它们）。
 - Review status: 2026-11-15 复查上游、许可证与维护状态。届时若上游 v1.1 真的放出数据层，
   重读上面"已知未收"那四条的可行性；不要在没有账本能力的前提下把它的 closeout 契约搬过来。
+
+## 2026-10 收敛：仓内只保留 catalog 正文
+
+本仓不再保存上游全量快照。除 catalog 命中的 1 颗正文、LICENSE 与本文件之外，以下内容按 2026-10 政策**从仓里删除**（不是仅从包里排除）；重新按上面所钉 commit 覆盖快照时按本节过滤。
+
+- **非 catalog skill（3 颗）**：`concept-research`、`industry-research`、`stock-research`。
+- **上游仓库材料**：`.npmignore`、`CHANGELOG.md`、`README.md`、`docs`、`examples`、`scripts`、`template`。
+- 许可与署名不受影响：`LICENSE` 原件、所钉 commit 与作者信息仍在上面各节；公开许可只要求保留版权与许可声明。

package/selected-skills/quant-agent-skills/skills/cross-market-listing-choice/SKILL.md

@@ -30,7 +30,7 @@ ADR 还是港股、已持有的那条线要不要换成另一条。
 
 这两句是这颗的核心，其余都是把它们变成可检查动作的细节。
 
-## 要拿的东西（一律走委派，本插件从不让分析角色自己伸手取数）
+## Evidence request
 
 向宿主的数据面要，一次把三项写全，缺哪一项就点名哪一项并停止依赖它的判断：
 

package/selected-skills/routing.js

@@ -597,7 +597,7 @@ export const SELECTED_SKILL_ROUTING = Object.freeze({
       en: "Use when the task matches this method: ranks upcoming events by what they move, when, and how they transmit across related names, and flags positions hanging on the same event chain. US and overseas calendar conventions. Triggers on Catalyst Map, Trade discipline, Macro & industry, Monitoring, US equity.",
     }),
     requires: Object.freeze(["filings","macro"]),
-    overlaps: Object.freeze(["catalyst-calendar","china-catalyst-calendar","risk-warning-catalysts","thesis-tracker","china-thesis-tracker"]),
+    overlaps: Object.freeze(["catalyst-calendar","china-catalyst-calendar","risk-warning-catalysts","thesis-tracker","china-thesis-tracker","macro-event-analysis"]),
     canonicalFor: Object.freeze([]),
   }),
   "watchlist-review": Object.freeze({
@@ -740,6 +740,34 @@ export const SELECTED_SKILL_ROUTING = Object.freeze({
     overlaps: Object.freeze(["portfolio-review","portfolio-concentration","risk-stress-test","position-sizing","position-ladder","watchlist-review","execution-plan-check","risk-reward-sanity-check","post-trade-review","journal-pattern-analyzer"]),
     canonicalFor: Object.freeze([]),
   }),
+  "macro-event-analysis": Object.freeze({
+    market: Object.freeze(["us-stock"]),
+    stage: "pre-event",
+    methods: Object.freeze(["macro","risk"]),
+    roles: Object.freeze(["main"]),
+    primaryRole: "main",
+    useWhen: Object.freeze({
+      zh: "当用户要在未来一到两周的宏观事件里挑出真正影响决策的几项，并看清它们通过利率、汇率、商品与风险偏好传导到持仓时使用；关键词：宏观事件、事件风险、传导、时间窗。",
+      en: "Use when the task matches this method: filters the next one to two weeks of macro events down to the decision-relevant few and maps their transmission through rates, FX, commodities and risk appetite. Triggers on Macro Event Analysis, Trade discipline, Macro & industry, Monitoring, US equity.",
+    }),
+    requires: Object.freeze(["macro"]),
+    overlaps: Object.freeze(["catalyst-map","market-regime-analysis"]),
+    canonicalFor: Object.freeze(["global/pre-event/macro"]),
+  }),
+  "market-regime-analysis": Object.freeze({
+    market: Object.freeze(["us-stock"]),
+    stage: "monitoring",
+    methods: Object.freeze(["technical","risk"]),
+    roles: Object.freeze(["main"]),
+    primaryRole: "main",
+    useWhen: Object.freeze({
+      zh: "当用户要按趋势、波动率、广度与事件背景判定当前市场环境，并据此选择趋势跟随、均值回归还是防御时使用；关键词：市场环境、趋势、波动率、广度。",
+      en: "Use when the task matches this method: classifies the environment across trend, volatility, breadth and event backdrop, then argues whether trend-following, mean reversion or caution is the better default. Triggers on Market Regime Analysis, Trade discipline, Technicals, Risk, US equity.",
+    }),
+    requires: Object.freeze(["price_history","macro"]),
+    overlaps: Object.freeze(["macro-event-analysis"]),
+    canonicalFor: Object.freeze(["global/monitoring/regime"]),
+  }),
   "china-earnings-analysis": Object.freeze({
     market: Object.freeze(["a-share"]),
     stage: "post-event",

package/selected-skills/trading-skills/UPSTREAM.md

@@ -123,3 +123,13 @@
   美股时段与盘前盘后、FINRA／Fidelity／Schwab 口径的集中度阈值、EURUSD 与 CFD／杠杆工具），
   **不声称适用 A 股**，也不得直接当作 A 股规则；它提供的是过程纪律而不是信号或收益承诺，
   所有金融主张须回到原始证据核验。本文件与本仓的收录判断不构成对上游观点的背书。
+
+## 2026-10 收敛：仓内只保留 catalog 正文
+
+本仓不再保存上游全量快照。除 catalog 命中的 13 颗正文、LICENSE 与本文件之外，以下内容按 2026-10 政策**从仓里删除**（不是仅从包里排除）；重新按上面所钉 commit 覆盖快照时按本节过滤。
+
+- **非 catalog skill（6 颗）**：`earnings-preview`、`earnings-trade-prep`、`etoro`、`portfolio-risk-review`、`post-trade-debrief`、`pre-trade-check`。
+- **上游仓库材料**：`.npmignore`。
+- **catalog 目录内的附属材料**：`macro-event-analysis` 与 `market-regime-analysis` 的 7 份服务商接口页（`references/providers/` 与 `references/data-providers.md`），随两颗收录前先删掉；其余 `references/` 是运行时正文。
+- **纯公式脚本（1 份）**：移到 `selected-skills/reference-algorithms/trading-skills/`，git-only、不随包、不执行。
+- 许可与署名不受影响：`LICENSE` 原件、所钉 commit 与作者信息仍在上面各节；公开许可只要求保留版权与许可声明。

package/selected-skills/trading-skills/skills/catalyst-map/SKILL.md

@@ -5,6 +5,13 @@ description: Build a ranked map of the catalysts that could move a watchlist, th
 
 # Catalyst Map
 
+## Evidence request
+
+通过宿主数据契约请求：披露文件（公告、年报或招股书，带回出处与文档日期）、宏观序列（数值、期间与发布日期）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill when the user wants a practical map of what could move a watchlist, sector, theme, or portfolio over a defined window, not just a raw event list.
 
 This skill will not:

package/selected-skills/trading-skills/skills/evidence-gap-check/SKILL.md

@@ -5,6 +5,13 @@ description: Identify the most important missing facts, assumptions, and unresol
 
 # Evidence Gap Check
 
+## Evidence request
+
+通过宿主数据契约请求：使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill when the user has an idea, thesis, or watchlist candidate but does not yet know whether the available evidence is sufficient to justify acting.
 
 This skill will not:

package/selected-skills/trading-skills/skills/execution-plan-check/SKILL.md

@@ -5,6 +5,13 @@ description: Review whether a trade plan is operationally executable by checking
 
 # Execution Plan Check
 
+## Evidence request
+
+通过宿主数据契约请求：行情序列（复权口径与 as-of）、使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill after the thesis and trade structure make sense but before the user places the order, when the main question is whether the plan can be executed cleanly in the real market.
 
 This skill will not:

package/selected-skills/trading-skills/skills/journal-pattern-analyzer/SKILL.md

@@ -5,6 +5,13 @@ description: Use when the user has a trade journal or trade log and wants repeat
 
 # Journal Pattern Analyzer
 
+## Evidence request
+
+通过宿主数据契约请求：持仓与成本（使用者提供）、使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill when the user has a set of trades, journal notes, or review entries and wants to know what patterns are repeating across them.
 
 This skill will not:

package/selected-skills/trading-skills/skills/macro-event-analysis/SKILL.md

@@ -41,18 +41,15 @@ Helpful but optional:
 - specific holdings or watchlist names
 - events the user already knows are important
 
-Use the user's materials first: pasted event lists, screenshots, platform notes, transcripts, watchlists, or provider details already mentioned in the conversation.
+Use the user's materials first: pasted event lists, screenshots, platform notes, transcripts, watchlists, or calendar details already mentioned in the conversation.
 
-## If critical data is missing
+## Evidence request
 
-If the user already gave enough event information to analyze, do not fetch anything.
+If the user already gave enough event information to analyze, do not request anything more.
 
-If the timing or event slate is still incomplete:
+If the timing or event slate is still incomplete, request an evidence pack through the host data contract: the market or exposure, the requested window, the countries / central banks / releases that matter, and the fields needed — event name, scheduled time in the user's timezone, prior and consensus values where published, and the source venue recorded on each row. Do not ask for a data vendor or credentials: the host decides where evidence comes from and returns it with provenance.
 
-- check whether the conversation already points to a supported provider or includes usable access details
-- if the user already indicated `FMP` or `TradingEconomics`, use [references/providers/fmp.md](references/providers/fmp.md) or [references/providers/tradingeconomics.md](references/providers/tradingeconomics.md) directly
-- otherwise consult [references/data-providers.md](references/data-providers.md) and ask which supported provider they want to use
-- once the missing event details are gathered, continue the analysis and disclose the source used
+When the missing event details arrive, continue the analysis and disclose the source recorded on each row. If a required field does not come back, name it and leave that part of the slate empty instead of estimating it.
 
 ## Analysis process
 
@@ -107,7 +104,7 @@ Always return:
 - the main transmission channels to watch
 - timing notes, including clustered or overnight risk
 - a short preparation brief: what to monitor, what to avoid assuming, and where caution is warranted
-- explicit source, freshness, and caveats when provider-based data is used
+- explicit source, freshness, and caveats when retrieved data is used
 
 ## Best practices
 

package/selected-skills/trading-skills/skills/macro-event-analysis/references/interpretation-guide.md

@@ -12,7 +12,7 @@ Focus on event risk, not prediction.
 ## What not to claim
 
 - that an event will move markets in a specific direction
-- that the provider has full global coverage
+- that the returned calendar has full global coverage
 - that stale timestamps are harmless
 
 ## Reporting guidance

package/selected-skills/trading-skills/skills/market-regime-analysis/SKILL.md

@@ -39,18 +39,15 @@ Ask for observations about:
 
 If the user is vague, ask for the minimum evidence needed instead of forcing a fake label.
 
-Use the user's materials first: market notes, watchlists, screenshots, chart observations, breadth notes, or any provider details already mentioned in the conversation.
+Use the user's materials first: market notes, watchlists, screenshots, chart observations, breadth notes, or any market data already mentioned in the conversation.
 
-## If critical data is missing
+## Evidence request
 
-If the user already supplied enough evidence to classify the environment, do not fetch anything.
+If the user already supplied enough evidence to classify the environment, do not request anything more.
 
-If trend, volatility, breadth, or event context is still too thin:
+If trend, volatility, breadth, or event context is still too thin, request an evidence pack through the host data contract: the market or index, the timeframe, and the observations needed — recent price history, realized range or volatility, breadth or participation measures, leadership concentration, and the macro / earnings calendar for the window. Do not ask for a data vendor or credentials: the host decides where evidence comes from and returns it with provenance.
 
-- check whether the user already named a supported provider or already shared usable access details
-- if they already indicated `FMP`, `TradingEconomics`, or `Polygon`, use [references/providers/fmp.md](references/providers/fmp.md), [references/providers/tradingeconomics.md](references/providers/tradingeconomics.md), or [references/providers/polygon.md](references/providers/polygon.md) directly
-- otherwise consult [references/data-providers.md](references/data-providers.md) and ask which supported provider they want to use
-- once the missing context is gathered, continue the regime analysis and disclose the source used
+Use the user's materials first: market notes, watchlists, screenshots, chart observations. If a required field does not come back, name it and classify conservatively on what is present instead of filling the gap from memory.
 
 ## Analysis process
 

package/selected-skills/trading-skills/skills/portfolio-concentration/SKILL.md

@@ -5,6 +5,13 @@ description: Evaluate whether a portfolio, account, or planned position is too d
 
 # Portfolio Concentration
 
+## Evidence request
+
+通过宿主数据契约请求：持仓与成本（使用者提供）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill before adding risk when the user needs to know whether a new or existing position would make the portfolio too dependent on a small set of holdings or shared drivers.
 
 This skill will not:

package/selected-skills/trading-skills/skills/position-management/SKILL.md

@@ -5,6 +5,13 @@ description: Review an open position and decide whether to hold, trim, tighten r
 
 # Position Management
 
+## Evidence request
+
+通过宿主数据契约请求：持仓与成本（使用者提供）、披露文件（公告、年报或招股书，带回出处与文档日期）、行情序列（复权口径与 as-of）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill after a position is open when the user needs a disciplined decision about holding, trimming, tightening risk, or exiting without drifting into improvisation.
 
 This skill will not:

package/selected-skills/trading-skills/skills/position-sizing/SKILL.md

@@ -5,6 +5,13 @@ description: Use when the user needs a conservative position size from account e
 
 # Position Sizing
 
+## Evidence request
+
+通过宿主数据契约请求：持仓与成本（使用者提供）、行情序列（复权口径与 as-of）、使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill before entering a trade when you need a defensible size instead of a gut-feel size.
 
 This skill will not:

package/selected-skills/trading-skills/skills/post-trade-review/SKILL.md

@@ -5,6 +5,13 @@ description: Guide a disciplined post-trade review across thesis quality, setup
 
 # Post Trade Review
 
+## Evidence request
+
+通过宿主数据契约请求：使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill after a trade closes or after a meaningful trade sequence when you want an honest process review.
 
 The objective is to improve repeatability, not to rewrite history.

package/selected-skills/trading-skills/skills/risk-reward-sanity-check/SKILL.md

@@ -5,6 +5,13 @@ description: Use when the user wants to test whether a proposed entry, stop, and
 
 # Risk Reward Sanity Check
 
+## Evidence request
+
+通过宿主数据契约请求：行情序列（复权口径与 as-of）、使用者提供的输入（成交、持仓或论点记录）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill when a trade idea exists but you want to inspect whether the structure makes sense before committing capital.
 
 This skill does not predict whether the trade will work. It checks whether the plan is internally coherent.

package/selected-skills/trading-skills/skills/thesis-validation/SKILL.md

@@ -5,6 +5,13 @@ description: Pressure-test a trade or investment thesis by clarifying the core c
 
 # Thesis Validation
 
+## Evidence request
+
+通过宿主数据契约请求：使用者提供的输入（成交、持仓或论点记录）、披露文件（公告、年报或招股书，带回出处与文档日期）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill before trade construction when the user has an idea, narrative, or directional lean but needs to know whether the thesis is clear enough to trust.
 
 This skill will not:

package/selected-skills/trading-skills/skills/watchlist-review/SKILL.md

@@ -5,6 +5,13 @@ description: Review a watchlist and rank which names deserve active attention, b
 
 # Watchlist Review
 
+## Evidence request
+
+通过宿主数据契约请求：持仓与成本（使用者提供）、行情序列（复权口径与 as-of）。
+
+缺哪个字段就点名并停止该字段下的分析；用户粘贴的材料标 `pasted by user`。
+
+
 Use this skill when the user already has a list of names, sectors, or themes and needs to narrow it into a smaller set of names worth preparing for, not just collecting.
 
 This skill will not:

package/watchlist/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@v587d/watchlist",
-  "version": "2.10.2",
+  "version": "2.11.0",
   "description": "自动模式的用户自选股（宿主级清单 + Fuyao 报价快照 + /capital-watchlist 路由 + 输入框菜单弹窗）：由 @v587d/capital-generation 随包携带并作为 host 平面行挂载。",
   "license": "MIT",
   "type": "module",