@hedgehog-finance/hedgehog-plugin 1.0.21 → 1.0.22

package/src/features/stockAnalysis/tools.ts

@@ -1,80 +1,291 @@
 import { randomUUID } from "node:crypto";
 import { getDB } from "../../core/database.js";
+import { CHART_OUTPUT_GUIDANCE, ensureChartPlaceholdersInBody } from "../chartOutput.js";
 import {
 	ArticleAiAnalysis,
+	BuildStockAiAnalysisMessageParams,
+	BuildStockAiAnalysisMessageParamsSchema,
 	GetArticleAiAnalysisParamsSchema,
+	GetStockAiAnalysisDetailParamsSchema,
 	GetStockAiAnalysisParamsSchema,
+	QueryArticleAiAnalysisHistoryParamsSchema,
+	QueryStockAiAnalysisStocksParamsSchema,
+	SaveArticleDeepReasoningParamsSchema,
 	QueryStockAiAnalysisHistoryParamsSchema,
 	SaveArticleAiAnalysisParamsSchema,
 	SaveStockAiAnalysisParamsSchema,
-	StockAiAnalysis
+	StockAiAnalysis,
+	StockAiAnalysisStockSummary
 } from "./schema.js";
 
 interface RuntimeTool {
 	name: string;
+	label?: string;
 	description: string;
 	parameters: unknown;
 	registerTool?: boolean;
-	execute(params: unknown, ctx: { userId: string }): Promise<string>;
+	execute(params: unknown, ctx?: { userId: string }): Promise<string>;
 }
 
-export function normalizeStockCode(stockCode: string): string {
-	return stockCode.trim().toUpperCase().replace(/\.SS$/i, ".SH");
+const STOCK_AI_ANALYSIS_SKILL = "hedgehog-stock-research";
+
+const BuildStockAiAnalysisMessageAgentToolSchema = {
+	type: "object",
+	additionalProperties: false,
+	required: ["stock_code", "stock_name"],
+	properties: {
+		stock_code: { type: "string", description: "股票代码" },
+		stock_name: { type: "string", description: "股票名称" },
+		market: { type: "string", description: "市场类型，默认 CN" }
+	}
+};
+
+export function normalizeStockCode(stock_code: string): string {
+	return stock_code.trim().toUpperCase().replace(/\.SS$/i, ".SH");
+}
+
+function resolveToolUserId(ctx?: { userId: string }): string {
+	return ctx?.userId || "default";
+}
+
+function buildStockAiAnalysisContent(args: BuildStockAiAnalysisMessageParams): string {
+	return [
+		`分析一下${args.stock_name}股票`,
+		"",
+		`股票代码：${normalizeStockCode(args.stock_code)}`,
+		`股票名称：${args.stock_name}`,
+		`市场：${args.market}`
+	].join("\n");
+}
+
+function buildStockAiAnalysisMessage(args: BuildStockAiAnalysisMessageParams): string {
+	const stock_code = normalizeStockCode(args.stock_code);
+	const buildGeneratingSaveParams = () => JSON.stringify({
+		stock_code,
+		stock_name: args.stock_name,
+		market: args.market,
+		status: "generating",
+		content: ""
+	});
+	const buildFinalSaveParams = (status: "completed" | "failed") => JSON.stringify({
+		stock_code,
+		stock_name: args.stock_name,
+		market: args.market,
+		status,
+		content: "..."
+	});
+	return JSON.stringify({
+		cw_system_prompt: [
+			`开始分析前必须先调用 save_stock_ai_analysis，参数为 ${buildGeneratingSaveParams()}；如果返回 skipped=true，必须立即停止，不要继续调用 skill 或生成正文。`,
+			`必须触发并使用 ${STOCK_AI_ANALYSIS_SKILL} skill 生成个股 AI 分析报告。`,
+			`生成成功后必须调用 save_stock_ai_analysis，参数为 ${buildFinalSaveParams("completed")}，content 传最终完整个股分析正文。`,
+			`生成失败后必须调用 save_stock_ai_analysis，参数为 ${buildFinalSaveParams("failed")}，content 存放完整错误信息。`,
+			"如果最终内容包含 [图表数据]，正文必须已经包含所有对应图表占位符。"
+		].join("\n"),
+		cw_context: JSON.stringify({
+			stock_code,
+			stock_name: args.stock_name,
+			market: args.market,
+			saveMode: "tool"
+		}),
+		cw_market: args.market,
+		cw_stock_code: stock_code,
+		cw_stock_name: args.stock_name,
+		cw_content: buildStockAiAnalysisContent({ ...args, stock_code }),
+		cw_output: [
+			`输出结构以 ${STOCK_AI_ANALYSIS_SKILL} skill 的交付模板为准。`,
+			CHART_OUTPUT_GUIDANCE
+		].join("\n")
+	});
 }
 
 function selectLatestStockAnalysis(
+	db: ReturnType<typeof getDB>,
+	stock_code: string,
+	market: string
+): StockAiAnalysis | undefined {
+	return db.prepare(`
+		SELECT id, stock_code, stock_name, market, status, content, createdAt, updatedAt
+		FROM stock_ai_analysis
+		WHERE stock_code = ? AND market = ?
+		ORDER BY updatedAt DESC, createdAt DESC
+		LIMIT 1
+	`).get(stock_code, market) as StockAiAnalysis | undefined;
+}
+
+function selectLatestGeneratingStockAnalysis(
 	db: ReturnType<typeof getDB>,
 	userId: string,
-	stockCode: string
+	stock_code: string,
+	market: string
+): StockAiAnalysis | undefined {
+	return db.prepare(`
+		SELECT id, stock_code, stock_name, market, status, content, createdAt, updatedAt
+		FROM stock_ai_analysis
+		WHERE userId = ? AND stock_code = ? AND market = ? AND status = 'generating'
+		ORDER BY updatedAt DESC, createdAt DESC
+		LIMIT 1
+	`).get(userId, stock_code, market) as StockAiAnalysis | undefined;
+}
+
+function selectStockAnalysisDetail(
+	db: ReturnType<typeof getDB>,
+	id: string
 ): StockAiAnalysis | undefined {
 	return db.prepare(`
-		SELECT id, stockCode, stockName, market, content, createdAt, updatedAt
+		SELECT id, stock_code, stock_name, market, status, content, createdAt, updatedAt
 		FROM stock_ai_analysis
-		WHERE userId = ? AND stockCode = ?
+		WHERE id = ?
 		ORDER BY updatedAt DESC, createdAt DESC
 		LIMIT 1
-	`).get(userId, stockCode) as StockAiAnalysis | undefined;
+	`).get(id) as StockAiAnalysis | undefined;
+}
+
+function queryStockAnalysisStocks(
+	db: ReturnType<typeof getDB>,
+	userId: string,
+	market: string,
+	page: number,
+	pageSize: number
+): { rows: StockAiAnalysisStockSummary[]; total: number } {
+	const offset = (page - 1) * pageSize;
+	const rows = db.prepare(`
+		WITH grouped AS (
+			SELECT
+				stock_code,
+				market,
+				COUNT(*) AS analysisCount,
+				MAX(updatedAt || '|' || createdAt || '|' || id) AS latestKey
+			FROM stock_ai_analysis
+			WHERE userId = ? AND market = ?
+			GROUP BY stock_code, market
+		)
+		SELECT
+			a.stock_code,
+			a.stock_name,
+			a.market,
+			a.id AS latestAnalysisId,
+			a.status AS latestStatus,
+			a.createdAt AS latestCreatedAt,
+			a.updatedAt AS latestUpdatedAt,
+			g.analysisCount
+		FROM grouped g
+		JOIN stock_ai_analysis a
+			ON a.stock_code = g.stock_code
+			AND a.market = g.market
+			AND (a.updatedAt || '|' || a.createdAt || '|' || a.id) = g.latestKey
+		WHERE a.userId = ?
+		ORDER BY a.updatedAt DESC, a.createdAt DESC
+		LIMIT ? OFFSET ?
+	`).all(userId, market, userId, pageSize, offset) as StockAiAnalysisStockSummary[];
+	const countRow = db.prepare(`
+		SELECT COUNT(*) AS total
+		FROM (
+			SELECT 1
+			FROM stock_ai_analysis
+			WHERE userId = ? AND market = ?
+			GROUP BY stock_code, market
+		)
+	`).get(userId, market) as { total: number };
+	return { rows, total: countRow.total || 0 };
 }
 
 export function saveStockAiAnalysisRecord(
 	db: ReturnType<typeof getDB>,
 	userId: string,
 	args: {
-		stockCode: string;
-		stockName: string;
+		stock_code: string;
+		stock_name?: string;
 		market: string;
 		content: string;
+		status?: string;
 	}
 ): StockAiAnalysis {
-	const stockCode = normalizeStockCode(args.stockCode);
+	const stock_code = normalizeStockCode(args.stock_code);
+	const status = args.status || "completed";
+	const stock_name = args.stock_name?.trim() || stock_code;
+	const content = status === "completed" ? ensureChartPlaceholdersInBody(args.content) : args.content.trim();
 	const id = randomUUID();
 
+	const generating = status === "generating"
+		? undefined
+		: selectLatestGeneratingStockAnalysis(db, userId, stock_code, args.market);
+	if (generating) {
+		db.prepare(`
+			UPDATE stock_ai_analysis
+			SET status = ?,
+				content = ?,
+				updatedAt = STRFTIME('%Y-%m-%dT%H:%M:%fZ', 'NOW')
+			WHERE id = ? AND userId = ?
+		`).run(status, content, generating.id, userId);
+		return selectStockAnalysisDetail(db, generating.id) as StockAiAnalysis;
+	}
+
 	db.prepare(`
-		INSERT INTO stock_ai_analysis (id, userId, stockCode, stockName, market, content)
-		VALUES (?, ?, ?, ?, ?, ?)
-	`).run(id, userId, stockCode, args.stockName, args.market, args.content);
+		INSERT INTO stock_ai_analysis (id, userId, stock_code, stock_name, market, status, content)
+		VALUES (?, ?, ?, ?, ?, ?, ?)
+	`).run(id, userId, stock_code, stock_name, args.market, status, content);
 
 	return db.prepare(`
-		SELECT id, stockCode, stockName, market, content, createdAt, updatedAt
+		SELECT id, stock_code, stock_name, market, status, content, createdAt, updatedAt
 		FROM stock_ai_analysis
 		WHERE userId = ? AND id = ?
 	`).get(userId, id) as StockAiAnalysis;
 }
 
+function tableForArticleAnalysis(analysisType: ArticleAiAnalysis["analysisType"]): string {
+	return analysisType === "verification" ? "news_fact_check_analysis" : "news_deep_reasoning_analysis";
+}
+
 function selectLatestArticleAnalysis(
 	db: ReturnType<typeof getDB>,
 	userId: string,
 	sourceId: string,
 	analysisType: ArticleAiAnalysis["analysisType"],
-	market: string
+	market: string = "CN"
+): ArticleAiAnalysis | undefined {
+	const table = tableForArticleAnalysis(analysisType);
+	if (analysisType === "deduction") {
+		return db.prepare(`
+			SELECT id, sourceId, ? AS analysisType, sourceTitle, market, status, content, createdAt, updatedAt
+			FROM ${table}
+			WHERE userId = ? AND sourceId = ? AND market = ?
+			ORDER BY updatedAt DESC, createdAt DESC
+			LIMIT 1
+		`).get(analysisType, userId, sourceId, market) as ArticleAiAnalysis | undefined;
+	}
+	return db.prepare(`
+		SELECT id, sourceId, ? AS analysisType, sourceTitle, status, content, createdAt, updatedAt
+		FROM ${table}
+		WHERE userId = ? AND sourceId = ?
+		ORDER BY updatedAt DESC, createdAt DESC
+		LIMIT 1
+	`).get(analysisType, userId, sourceId) as ArticleAiAnalysis | undefined;
+}
+
+function selectLatestGeneratingArticleAnalysis(
+	db: ReturnType<typeof getDB>,
+	userId: string,
+	sourceId: string,
+	analysisType: ArticleAiAnalysis["analysisType"]
 ): ArticleAiAnalysis | undefined {
+	const table = tableForArticleAnalysis(analysisType);
+	if (analysisType === "deduction") {
+		return db.prepare(`
+			SELECT id, sourceId, ? AS analysisType, sourceTitle, market, status, content, createdAt, updatedAt
+			FROM ${table}
+			WHERE userId = ? AND sourceId = ? AND status = 'generating'
+			ORDER BY updatedAt DESC, createdAt DESC
+			LIMIT 1
+		`).get(analysisType, userId, sourceId) as ArticleAiAnalysis | undefined;
+	}
 	return db.prepare(`
-		SELECT id, sourceId, analysisType, market, content, createdAt, updatedAt
-		FROM article_ai_analysis
-		WHERE userId = ? AND sourceId = ? AND analysisType = ? AND market = ?
+		SELECT id, sourceId, ? AS analysisType, sourceTitle, status, content, createdAt, updatedAt
+		FROM ${table}
+		WHERE userId = ? AND sourceId = ? AND status = 'generating'
 		ORDER BY updatedAt DESC, createdAt DESC
 		LIMIT 1
-	`).get(userId, sourceId, analysisType, market) as ArticleAiAnalysis | undefined;
+	`).get(analysisType, userId, sourceId) as ArticleAiAnalysis | undefined;
 }
 
 function saveArticleAiAnalysisRecord(
@@ -82,63 +293,100 @@ function saveArticleAiAnalysisRecord(
 	userId: string,
 	args: {
 		sourceId: string;
-		analysisType: ArticleAiAnalysis["analysisType"];
-		market: string;
+		sourceTitle: string;
 		content: string;
+		status: string;
 	}
 ): ArticleAiAnalysis {
 	const id = randomUUID();
 
 	db.prepare(`
-		INSERT INTO article_ai_analysis (id, sourceId, userId, analysisType, market, content)
+		INSERT INTO news_fact_check_analysis (id, sourceId, sourceTitle, userId, status, content)
 		VALUES (?, ?, ?, ?, ?, ?)
-		ON CONFLICT(sourceId, userId, analysisType, market) DO UPDATE SET
+		ON CONFLICT(sourceId, userId) DO UPDATE SET
+			sourceTitle = CASE WHEN excluded.sourceTitle != '' THEN excluded.sourceTitle ELSE news_fact_check_analysis.sourceTitle END,
+			status = excluded.status,
+			content = excluded.content,
+			updatedAt = STRFTIME('%Y-%m-%dT%H:%M:%fZ', 'NOW')
+	`).run(id, args.sourceId, args.sourceTitle, userId, args.status, args.content);
+
+	return db.prepare(`
+		SELECT id, sourceId, 'verification' AS analysisType, sourceTitle, status, content, createdAt, updatedAt
+		FROM news_fact_check_analysis
+		WHERE userId = ? AND sourceId = ?
+	`).get(userId, args.sourceId) as ArticleAiAnalysis;
+}
+
+function saveArticleDeepReasoningRecord(
+	db: ReturnType<typeof getDB>,
+	userId: string,
+	args: {
+		sourceId: string;
+		sourceTitle: string;
+		market: string;
+		content: string;
+		status: string;
+	}
+): ArticleAiAnalysis {
+	const id = randomUUID();
+
+	db.prepare(`
+		INSERT INTO news_deep_reasoning_analysis (id, sourceId, sourceTitle, userId, market, status, content)
+		VALUES (?, ?, ?, ?, ?, ?, ?)
+		ON CONFLICT(sourceId, userId, market) DO UPDATE SET
+			sourceTitle = CASE WHEN excluded.sourceTitle != '' THEN excluded.sourceTitle ELSE news_deep_reasoning_analysis.sourceTitle END,
+			status = excluded.status,
 			content = excluded.content,
 			updatedAt = STRFTIME('%Y-%m-%dT%H:%M:%fZ', 'NOW')
-	`).run(id, args.sourceId, userId, args.analysisType, args.market, args.content);
+	`).run(id, args.sourceId, args.sourceTitle, userId, args.market, args.status, args.content);
 
 	return db.prepare(`
-		SELECT id, sourceId, analysisType, market, content, createdAt, updatedAt
-		FROM article_ai_analysis
-		WHERE userId = ? AND sourceId = ? AND analysisType = ? AND market = ?
-	`).get(userId, args.sourceId, args.analysisType, args.market) as ArticleAiAnalysis;
+		SELECT id, sourceId, 'deduction' AS analysisType, sourceTitle, market, status, content, createdAt, updatedAt
+		FROM news_deep_reasoning_analysis
+		WHERE userId = ? AND sourceId = ? AND market = ?
+	`).get(userId, args.sourceId, args.market) as ArticleAiAnalysis;
 }
 
 export const stockAnalysisTools: Record<string, RuntimeTool> = {
 	get_stock_ai_analysis: {
 		name: "get_stock_ai_analysis",
-		description: "读取股票 AI 分析的最新一条历史记录；不触发模型分析。",
+		description: "查询指定股票最新一条 AI 分析详情。该接口仅读取已持久化的分析结果，返回完整 content，不触发新的模型生成流程。",
 		parameters: GetStockAiAnalysisParamsSchema,
 		registerTool: false,
 		async execute(params, ctx) {
 			const args = GetStockAiAnalysisParamsSchema.parse(params);
 			const db = getDB();
-			const data = selectLatestStockAnalysis(db, ctx.userId, normalizeStockCode(args.stockCode));
+			const data = selectLatestStockAnalysis(db, normalizeStockCode(args.stock_code), args.market);
 			return JSON.stringify({ success: true, data: data || null });
 		}
 	},
 	query_stock_ai_analysis_history: {
 		name: "query_stock_ai_analysis_history",
-		description: "分页读取股票 AI 分析历史记录；不触发模型分析。",
+		description: "分页查询股票 AI 分析记录列表。支持按股票代码和市场过滤，返回记录标识、股票信息、状态和时间字段；列表结果不包含 content，详情内容请使用详情查询接口获取。",
 		parameters: QueryStockAiAnalysisHistoryParamsSchema,
 		registerTool: false,
 		async execute(params, ctx) {
-			const args = QueryStockAiAnalysisHistoryParamsSchema.parse(params);
+			const args = QueryStockAiAnalysisHistoryParamsSchema.parse(params ?? {});
 			const db = getDB();
-			const stockCode = normalizeStockCode(args.stockCode);
 			const offset = (args.page - 1) * args.pageSize;
+			const conditions = ["userId = ?", "market = ?"];
+			const queryParams: unknown[] = [resolveToolUserId(ctx), args.market];
+			if (args.stock_code) {
+				conditions.push("stock_code = ?");
+				queryParams.push(normalizeStockCode(args.stock_code));
+			}
 			const rows = db.prepare(`
-				SELECT id, stockCode, stockName, market, content, createdAt, updatedAt
+				SELECT id, stock_code, stock_name, market, status, createdAt, updatedAt
 				FROM stock_ai_analysis
-				WHERE userId = ? AND stockCode = ? AND market = ?
+				WHERE ${conditions.join(" AND ")}
 				ORDER BY updatedAt DESC, createdAt DESC
 				LIMIT ? OFFSET ?
-			`).all(ctx.userId, stockCode, args.market, args.pageSize, offset) as StockAiAnalysis[];
+			`).all(...queryParams, args.pageSize, offset);
 			const countRow = db.prepare(`
 				SELECT COUNT(*) AS total
 				FROM stock_ai_analysis
-				WHERE userId = ? AND stockCode = ? AND market = ?
-			`).get(ctx.userId, stockCode, args.market) as { total: number };
+				WHERE ${conditions.join(" AND ")}
+			`).get(...queryParams) as { total: number };
 			const total = countRow.total || 0;
 
 			return JSON.stringify({
@@ -153,39 +401,189 @@ export const stockAnalysisTools: Record<string, RuntimeTool> = {
 			});
 		}
 	},
+	query_stock_ai_analysis_stocks: {
+		name: "query_stock_ai_analysis_stocks",
+		description: "分页查询当前用户所有已经产生过个股 AI 分析记录的股票列表。按股票代码和市场去重，返回每只股票最近一次分析记录 ID、最近状态、最近更新时间和累计分析次数；不返回分析正文 content。",
+		parameters: QueryStockAiAnalysisStocksParamsSchema,
+		registerTool: false,
+		async execute(params, ctx) {
+			const args = QueryStockAiAnalysisStocksParamsSchema.parse(params ?? {});
+			const db = getDB();
+			const { rows, total } = queryStockAnalysisStocks(db, resolveToolUserId(ctx), args.market, args.page, args.pageSize);
+			return JSON.stringify({
+				success: true,
+				data: rows,
+				pagination: {
+					page: args.page,
+					pageSize: args.pageSize,
+					total,
+					totalPages: Math.ceil(total / args.pageSize)
+				}
+			});
+		}
+	},
+	get_stock_ai_analysis_detail: {
+		name: "get_stock_ai_analysis_detail",
+		description: "根据分析记录 ID 查询股票 AI 分析详情。返回指定记录的完整分析内容 content 及元数据，用于详情页展示或历史记录回放。",
+		parameters: GetStockAiAnalysisDetailParamsSchema,
+		registerTool: false,
+		async execute(params, ctx) {
+			const args = GetStockAiAnalysisDetailParamsSchema.parse(params);
+			const db = getDB();
+			const data = selectStockAnalysisDetail(db, args.id);
+			return JSON.stringify({ success: true, data: data || null });
+		}
+	},
+	build_stock_ai_analysis_message: {
+		name: "build_stock_ai_analysis_message",
+		label: "构建个股分析消息",
+		description: "根据股票代码、名称和市场构建用于主动 RPC 发起 Agent 个股 AI 分析任务的标准消息。该工具只返回提示词消息体，不触发定时任务，也不保存分析结果。",
+		parameters: BuildStockAiAnalysisMessageAgentToolSchema,
+		registerTool: false,
+		async execute(params, ctx) {
+			const args = BuildStockAiAnalysisMessageParamsSchema.parse(params);
+			const db = getDB();
+			const userId = resolveToolUserId(ctx);
+			const stock_code = normalizeStockCode(args.stock_code);
+			const generating = selectLatestGeneratingStockAnalysis(db, userId, stock_code, args.market);
+			if (generating) {
+				return JSON.stringify({
+					success: true,
+					skipped: true,
+					reason: "already_generating",
+					data: generating
+				});
+			}
+			const message = buildStockAiAnalysisMessage({ ...args, stock_code });
+			return JSON.stringify({
+				success: true,
+				data: {
+					message,
+					payload: JSON.parse(message),
+					stock_code,
+					saveParams: {
+						stock_code,
+						stock_name: args.stock_name,
+						market: args.market
+					},
+					skill: STOCK_AI_ANALYSIS_SKILL
+				}
+			});
+		}
+	},
 	save_stock_ai_analysis: {
 		name: "save_stock_ai_analysis",
-		description: "追加保存一条股票 AI 分析历史记录。",
+		description: [
+			"保存个股 AI 分析结果。生成前必须先以 status=generating、content=\"\" 调用，并传入 stock_code、stock_name、market；生成成功后以 status=completed 保存完整正文 content；生成失败后以 status=failed 保存完整错误信息。",
+			"个股分析的 cw_output 需要包含图表相关要求：",
+			CHART_OUTPUT_GUIDANCE
+		].join("\n"),
 		parameters: SaveStockAiAnalysisParamsSchema,
-		registerTool: false,
+		registerTool: true,
 		async execute(params, ctx) {
 			const args = SaveStockAiAnalysisParamsSchema.parse(params);
 			const db = getDB();
-			const data = saveStockAiAnalysisRecord(db, ctx.userId, args);
+			const userId = resolveToolUserId(ctx);
+			if (args.status === "generating") {
+				const stock_code = normalizeStockCode(args.stock_code);
+				const generating = selectLatestGeneratingStockAnalysis(db, userId, stock_code, args.market);
+				if (generating) {
+					return JSON.stringify({ success: true, skipped: true, reason: "already_generating", data: generating });
+				}
+			}
+			const data = saveStockAiAnalysisRecord(db, userId, args);
 			return JSON.stringify({ success: true, data });
 		}
 	},
 	get_article_ai_analysis: {
 		name: "get_article_ai_analysis",
-		description: "读取文章 AI 分析结果；支持信息求证与深度推演，不触发模型分析。",
+		description: "根据资讯来源 ID 查询文章 AI 分析详情。支持信息求证和深度推演两类结果，返回完整 content；该接口仅读取已持久化数据，不触发新的模型生成流程。",
 		parameters: GetArticleAiAnalysisParamsSchema,
 		registerTool: false,
 		async execute(params, ctx) {
 			const args = GetArticleAiAnalysisParamsSchema.parse(params);
 			const db = getDB();
-			const data = selectLatestArticleAnalysis(db, ctx.userId, args.id, args.analysisType, args.market);
+			const data = selectLatestArticleAnalysis(db, resolveToolUserId(ctx), args.sourceId, args.analysisType, args.market);
 			return JSON.stringify({ success: true, data: data || null });
 		}
 	},
-	save_article_ai_analysis: {
-		name: "save_article_ai_analysis",
-		description: "保存文章 AI 分析结果；支持信息求证与深度推演。",
-		parameters: SaveArticleAiAnalysisParamsSchema,
+	query_article_ai_analysis_history: {
+		name: "query_article_ai_analysis_history",
+		description: "分页查询文章 AI 分析记录列表。支持按分析类型过滤，返回记录标识、sourceId、状态和时间字段；列表结果不包含 content，详情内容请按 sourceId 查询。",
+		parameters: QueryArticleAiAnalysisHistoryParamsSchema,
 		registerTool: false,
+		async execute(params, ctx) {
+			const args = QueryArticleAiAnalysisHistoryParamsSchema.parse(params ?? {});
+			const db = getDB();
+			const userId = resolveToolUserId(ctx);
+			const table = tableForArticleAnalysis(args.analysisType);
+			const offset = (args.page - 1) * args.pageSize;
+			const marketCondition = args.analysisType === "deduction" ? " AND market = ?" : "";
+			const queryParams = args.analysisType === "deduction"
+				? [args.analysisType, userId, args.market, args.pageSize, offset]
+				: [args.analysisType, userId, args.pageSize, offset];
+			const rows = db.prepare(`
+				SELECT id, sourceId, ? AS analysisType, sourceTitle, ${args.analysisType === "deduction" ? "market," : ""} status, createdAt, updatedAt
+				FROM ${table}
+				WHERE userId = ?${marketCondition}
+				ORDER BY updatedAt DESC, createdAt DESC
+				LIMIT ? OFFSET ?
+			`).all(...queryParams);
+			const countParams = args.analysisType === "deduction" ? [userId, args.market] : [userId];
+			const countRow = db.prepare(`
+				SELECT COUNT(*) AS total
+				FROM ${table}
+				WHERE userId = ?${marketCondition}
+			`).get(...countParams) as { total: number };
+			const total = countRow.total || 0;
+
+			return JSON.stringify({
+				success: true,
+				data: rows,
+				pagination: {
+					page: args.page,
+					pageSize: args.pageSize,
+					total,
+					totalPages: Math.ceil(total / args.pageSize)
+				}
+			});
+		}
+	},
+	save_information_verification: {
+		name: "save_information_verification",
+		description: "保存新闻信息求证结果。生成前必须先以 status=generating、content=\"\" 调用，并传入 sourceId、sourceTitle；生成成功后以 status=completed 保存完整正文 content；生成失败后以 status=failed 保存完整错误信息。",
+		parameters: SaveArticleAiAnalysisParamsSchema,
+		registerTool: true,
 		async execute(params, ctx) {
 			const args = SaveArticleAiAnalysisParamsSchema.parse(params);
 			const db = getDB();
-			const data = saveArticleAiAnalysisRecord(db, ctx.userId, { ...args, sourceId: args.id });
+			const userId = resolveToolUserId(ctx);
+			if (args.status === "generating") {
+				const generating = selectLatestGeneratingArticleAnalysis(db, userId, args.sourceId, "verification");
+				if (generating) {
+					return JSON.stringify({ success: true, skipped: true, reason: "already_generating", data: generating });
+				}
+			}
+			const data = saveArticleAiAnalysisRecord(db, userId, args);
+			return JSON.stringify({ success: true, data });
+		}
+	},
+	save_article_deep_reasoning_analysis: {
+		name: "save_article_deep_reasoning_analysis",
+		description: "保存新闻深度推演结果。生成前必须先以 status=generating、content=\"\" 调用，并传入 sourceId、sourceTitle、sourceContent、market；生成成功后以 status=completed 保存完整正文 content；生成失败后以 status=failed 保存完整错误信息。",
+		parameters: SaveArticleDeepReasoningParamsSchema,
+		registerTool: true,
+		async execute(params, ctx) {
+			const args = SaveArticleDeepReasoningParamsSchema.parse(params);
+			const db = getDB();
+			const userId = resolveToolUserId(ctx);
+			if (args.status === "generating") {
+				const generating = selectLatestGeneratingArticleAnalysis(db, userId, args.sourceId, "deduction");
+				if (generating) {
+					return JSON.stringify({ success: true, skipped: true, reason: "already_generating", data: generating });
+				}
+			}
+			const data = saveArticleDeepReasoningRecord(db, userId, args);
 			return JSON.stringify({ success: true, data });
 		}
 	}

package/src/features/stockBasic/schema.ts

@@ -0,0 +1,33 @@
+import { z } from "zod";
+
+export const StockBasicItemSchema = z.object({
+	act_ent_type: z.string().optional().default(""),
+	act_name: z.string().optional().default(""),
+	area: z.string().optional().default(""),
+	cnspell: z.string().optional().default(""),
+	curr_type: z.string().optional().default(""),
+	enname: z.string().optional().default(""),
+	exchange: z.string().trim().min(1),
+	fullname: z.string().optional().default(""),
+	industry: z.string().optional().default(""),
+	is_hs: z.string().optional().default(""),
+	list_date: z.string().optional().default(""),
+	market: z.string().optional().default(""),
+	name: z.string().trim().min(1),
+	stock_code: z.string().trim().min(1),
+	symbol: z.string().trim().min(1)
+});
+export type StockBasicItem = z.infer<typeof StockBasicItemSchema>;
+
+export const SyncStockBasicParamsSchema = z.object({
+	stocks: z.array(StockBasicItemSchema).min(1)
+});
+export type SyncStockBasicParams = z.infer<typeof SyncStockBasicParamsSchema>;
+
+export const GetStockBasicListParamsSchema = z.object({}).nullish();
+export type GetStockBasicListParams = z.infer<typeof GetStockBasicListParamsSchema>;
+
+export const GetStockBasicInfoParamsSchema = z.object({
+	stock_code: z.string().trim().min(1).describe("股票代码")
+});
+export type GetStockBasicInfoParams = z.infer<typeof GetStockBasicInfoParamsSchema>;