Stock Picker Orchestrator — Stock Picker 编排器

Stock Picker 编排器

Use this 技能 to coordinate the full analysis 系统 from user intent to final recommendation framing.

Purpose Convert user 请求 into the right analysis 流水线. Control bud获取: vnstock API calls, breadth of news scrAPIng, depth of valuation work. Produce transparent 输出s: what was fetched, assumptions, confidence, gaps. Scope boundary: this 技能 coordinates other 技能s and does not replace their domAIn-specific 记录ic. 技能 graph (preferred dependencies) vnstock-free-expert for structured market/fundamental data. nso-macro-监控 for Vietnam macro snapshot. us-macro-news-监控 for global macro spillover 签名als. vn-market-news-监控 for domestic market narrative. equity-valuation-框架 for decision-grade valuation and 报告 standard. portfolio-risk-管理器 for IPS mini + position sizing + risk triggers (no-margin). Trigger conditions "Find best stock(s)" "Screen this sector" "Analyze ticker X deeply" "How do macro/news affect these stocks" "Value this stock like a professional" First step: intent classification

Classify user 请求 into one of these modes:

Single-Ticker Deep Dive Multi-Ticker/Universe Screening Macro/News-Led Investigation Portfolio Refresh

If ambiguous, choose the most conservative high-签名al mode and note assumption.

Execution 工作流 (ordered) 解析 user intent and select one routing mode. 设置 bud获取 pre设置 (Light, Standard, Deep) and hard 请求 limits. 执行 required up流 技能s for the chosen 路由. 验证 intermediate 输出s for freshness, completeness, and conflicts. 运行 valuation layer only at the required depth. 聚合 confidence across 模块s using the 分享d rubric. Return 输出 using the mandatory 输出 contract. Bud获取 policy (required)

Define and enforce bud获取 at 启动:

API bud获取: max vnstock calls News bud获取: max headlines/articles per source Valuation depth: quick multiples vs full DCF

Default safe pre设置s:

Light: 20-40 vnstock calls, headlines-only news, quick valuation Standard: 40-120 calls, mixed headlines + selected deep reads, scenario valuation Deep: 120+ calls, full 上下文 package, full valuation + sensitivity

Prefer free-tier-safe pacing when using vnstock.

Free-tier bud获取 m应用ing (required)

Use these hard limits for vnstock 运行s:

Guest/no API key: max 20 请求s/min (recommended pacing >= 3.2s/请求). Community API key: max 60 请求s/min (recommended pacing >= 1.1s/请求; keep 3.2s/请求 if unstable).

Policy actions:

Estimate call count before execution and choose the smallest viable pre设置. If estimated calls exceed current bud获取, reduce scope (smaller universe or fewer 模块s). Reuse 缓存d artifacts before making new 请求s. 停止 scope expansion when remAIning call bud获取 < 10% and 报告 partial 结果s. Routing 记录ic A) Single ticker 请求

Priority: depth over breadth. 流水线:

vnstock-free-expert fetch financials + price behavior. Optional macro/news 上下文 if user asks or risk is macro-sensitive. equity-valuation-框架 full thesis + valuation + risks. B) Multi-ticker/sector screening

Priority: breadth first, then depth on fina列出s. 流水线:

vnstock-free-expert broad screener/ranking. Select top candidates by objective criteria. 运行 quick valuation layer on short列出. Deep valuation only for top 1-3 names. C) Macro/news-led 请求

Priority: 上下文 first, valuation second. 流水线:

nso-macro-监控 + us-macro-news-监控 + vn-market-news-监控. Map exposures to sectors/tickers. 运行 quick vnstock 验证 on impacted names. If needed, 运行 equity-valuation-框架 for decision-critical names. D) Portfolio refresh

Priority: risk control + 监控ing triggers + sizing discipline. 流水线:

Re-score holdings and benchmark agAInst alternatives. Macro/news stress overlay. 运行 equity-valuation-框架 at least quick depth on key holdings/watch列出. 运行 portfolio-risk-管理器 to produce IPS mini + position sizing policy + per-ticker triggers/in验证. Flag rebalance candidates with confidence and data gaps. Mandatory 输出 contract

Always include these sections in final 响应:

What Was Fetched Data sources used, date/time, and coverage. 流水线 Chosen Why this 路由 was selected for current user intent. Assumptions Explicit assumptions on macro, valuation parameters, and data 质量. 结果s Ranked 输出s or thesis summary with concise evidence. Confidence and Gaps Confidence level + missing data + potential impact. Risk Flags Top risks and 监控ing triggers. Next-Step Options 2-3 practical follow-up actions (e.g., deepen 1 ticker, expand peer 设置, 更新 after next macro release). 分享d confidence rubric (required)

Use a unified confidence 输出 across 流水线 steps:

High: all critical 模块s complete with no material data blockers. Medium: one critical 模块 has partial gaps but overall conclusion remAIns stable. Low: key 模块(s) missing or conflicting evidence makes conclusion fragile.

Aggregation rule:

Compute