gangtise-openapi-cli 0.24.0 → 0.26.0

package/README.md

@@ -4,6 +4,38 @@
 
 ## Changelog
 
+### v0.26.0 — 2026-07-11
+
+**资金安全（重要）**
+- 13 个贵档端点（`one-pager` / `investment-logic` / `peer-comparison` / `research-outline` / `theme-tracking` / `management-discuss-*`×2 / `hot-topic` / `knowledge-batch` / `earnings-review get-id` / `viewpoint-debate get-id` / `concept-info` / `concept-securities`）改为 **no-replay 重试策略**：5xx / 超时 / `999999` 不再自动重放——实测（2026-07-11）平台按次计费且**缓存命中不豁免**，同参数重放每次都扣分；仅连接期错误（`ECONNREFUSED`/DNS，请求未发出）、429 限流和 token 自愈仍重试。便宜按条计费的 list 类维持原全量重试（失败响应没有数据行、不计费）
+- 连接失败 `ECONNREFUSED` / `UND_ERR_CONNECT_TIMEOUT` 纳入默认重试范围（此前这两类不重试）
+
+**文件安全**
+- 所有 `--output` 落盘（导出、流式下载主路径、签名 URL 跟随下载）改为原子写：先写同目录 `.part` 成功后 rename——重跑失败不再毁掉已有旧文件；顺带修掉中止路径上 `.part` 因流懒打开竞态残留的问题
+- 签名 URL 跟随下载改走 transport 层：遵守 `GANGTISE_TIMEOUT_MS`（此前裸 `fetch` 无超时，慢滴速 CDN 可无限挂起）、网络错误自动重试、跟随最多 3 跳重定向（undici 不自动跟随，超限/缺 `Location` 报错而非把跳转页存成文件）、`--verbose` 日志剥离签名 query 只留 origin+path
+- 自动命名去重后缀试尽 `-1`…`-99` 仍冲突时报错，不再静默覆盖最早的文件
+
+**修复 / 加固**
+- 下载重定向超过 3 跳或缺 `Location` 时报错，不再把跳转页 HTML 当文件内容保存
+- 损坏的 gzip 响应包装为带请求上下文的 `ApiError`（此前抛裸 zlib `Z_DATA_ERROR`，与请求无关联且不可定位）
+- `alternative edb-search --limit` ≤200、`indicator search --limit` ≤100 本地上限校验——实测服务端对超限值静默截断（201→200、101→100），与 v0.25.0 的 `--top` 同类同修法
+
+### v0.25.0 — 2026-07-10
+
+**新增接口（4）**
+- `insight qa list` — 投资者问答 QA：按证券提取互动平台 / 电话会议 / 调研纪要的提问与回答；`--security-code`（必填）、`--source`（`conference`/`interactive`/`survey`）、`--question-category`（11 类，见 `insight.md`）、`--answer-important`（`1` 是 / `0` 否）、`--start-time`/`--end-time`（字符串直传）；自动翻页（单页上限 500）；0.1 积分/条
+- `insight report-image list` / `download` — 研报图表：按关键词搜索研报图片，返回 `chunkId` + 元数据（`--keyword` 必填、`--top` 默认 10 上限 20、`--source-id`、时间过滤；**免费**），再 `download --chunk-id` 下二进制原图（JPEG，0.1 积分/张）
+- `reference official-account-search` — 公众号 ID 搜索：输入公众号名 / 机构 / 关键字返回 `accountId`（喂 `insight official-account list --account-id`）；`--keyword`（必填）、`--category`（`listedCompany`/`broker`/`government`/`media`，可重复；未分类公众号 `category` 为 `null`，传 `--category` 会漏掉）、`--top`（默认 10 上限 10）；免费
+
+**变更**
+- `indicator search` / `cross-section` / `time-series` 市场范围从仅 A 股扩展至 A 股 / 港股 / 美股（服务端变更；CLI 早已支持 `--currency` 与多市场证券代码，无需改动）。⚠️ 美股代码用交易所后缀 `.O`(NASDAQ) / `.N`(NYSE)，**非 `.US`**——官方示例的 `AAPL.US` 查不到数据，实测须 `AAPL.O`
+
+**修复 / 加固**（承接上一批未单独发版的改动）
+- 分页 / 分片 `partial` 检测补全：`requestPaginated` 的短后续页、`MAX_PAGES` 上限、`total` 漂移、失败页四种场景统一触发 `partial`（退出码 3）——失败页独立成判定条件，避免超额返回的兄弟页把行数补满、掩盖失败页空洞；`quote` 全市场分片硬错后熔断、破损形状分片计入 `failedShards`
+- `--top` 本地上限校验（`report-image` / `knowledge-batch` ≤20，reference 六个搜索命令 ≤10）——实测服务端对超限值**静默截断**不报错，现在发请求前本地报错；`securities-search` / `institution-search` / `official-account-search` 的 `--category` 加本地白名单——实测服务端对拼错的分类**不报错**（securities-search 静默忽略过滤返回全类别、另两个静默返回空），拼写错误不再伪装成"无结果"（`insight qa` 的枚举服务端会报 `100003`，故不做本地白名单）
+- 错误码 `100003`（参数值非法）补充中文提示——服务端不指明是哪个参数，提示对照命令 `--help` 检查枚举参数拼写
+- undici `^7.16.0` → `^7.28.0`（修 keep-alive 队列污染 GHSA-35p6-xmwp-9g52），`engines.node` `>=20` → `>=20.18.1` 对齐 undici 实际最低要求
+
 ### v0.24.0 — 2026-07-07
 
 **新增**
@@ -44,75 +76,6 @@
 **文档 / Skill**（随 `/sync-skill` 分发，不影响 CLI 行为）
 - gangtise-openapi Agent Skill 经 fable5 审计 + 多轮 review 优化：官方积分计费速查表 + 高积分 pre-flight 闸门、AI 同步生成命令 `GANGTISE_TIMEOUT_MS=120000` 超时前置、大结果集 `--output` 落盘、异步 `--wait` 主路径、行业码口径收敛到单一权威、市值量纲实测（`qte_mkt_cptl` 仅 A 股 / 默认原始「元」/ `scale`+`currency`）等文档补全与消歧
 
-### v0.22.0 — 2026-07-02
-
-**行为变更（注意）**
-- ⚠️ 自动翻页接口省略 `--size` 现在一律拉全量（不再区分是否传时间范围）；需要只取前 N 条时请显式传 `--size N`。数据量未知时可先用 `--size 1` 从 stderr 的 `Total: N` 探明量级
-- 部分结果可机器识别：翻页页失败、K 线分片失败、或服务端提前短页但仍报告更大 `total` 时，结果会带 `partial: true`（页失败另有 `failedPages`，分片为 `failedShards`），非 json 行式输出仍只输出数据行，但进程退出码为 3
-
-**修复（鉴权 / 请求可靠性）**
-- Token 自愈覆盖服务端 `0000001008` 踢线失效，并能处理 HTTP 4xx 错误信封；`GANGTISE_TOKEN` + AK/SK 场景下环境 token 失效后不再反复回放旧 token
-- 并发请求同时遇到旧 token 失效时复用一次刷新结果；若刚拿到的新 token 本身被踢掉，则强制再次登录，避免"刚登录窗口期"误跳过刷新
-- 自动重试范围扩展到 429、DNS/网络临时错误与 undici 超时类错误；`GANGTISE_BASE_URL` 带路径前缀时 URL 拼接不再丢前缀
-
-**修复（下载 / 输出 / 数据正确性）**
-- 下载接口跟随最多 3 次 30x 跳转；跨域跳到对象存储签名 URL 时不携带 Authorization；服务端返回 `{url}` 且用户传 `--output` 时会真正下载文件，而不是把 URL 字符串写进文件
-- 自动文件名补齐清洗、截断与去重：服务端文件名、标题缓存名和 fallback 名都不会把 `/`、控制字符、过长中文名或重复标题变成路径/覆盖问题
-- `table`/`markdown` 输出清理控制字符、正确按 CJK 宽字符对齐，并转义 markdown 表头中的 `|`；CSV 输出转义表头、文件输出带 UTF-8 BOM，流式 CSV 遇全标量列表时回退到正常渲染而不是只写 BOM
-- `indicator search` / `cross-section` / `time-series` 的内层失败信封即使没有 `data` 字段也会抛出 `ApiError`，不再把"无权限/参数错误"渲染成成功结果
-- `--indicator-param` 等逗号列表支持全角逗号 `，`；日期型时间参数按本地零点解析，避免 `yyyy-MM-dd` 被当作 UTC 造成查询窗口偏移
-- `fundamental earning-forecast` 省略 `--start-date` 时按传入的 `--end-date` 往前一年计算，不再总是按今天往前一年
-- AI 异步 `--wait` 对 `410111` 终态失败只提示"不要重试"，超时才提示稍后用 check 命令查询；等待说明同步为最长约 5 分钟
-
-**CLI / 工程**
-- `raw call` 会在本地拒绝 JSON endpoint 的 `--query` 和 download endpoint 的 `--body`，避免静默丢参数；`--format` 在发请求前校验，格式拼错不再先消耗接口调用
-- `gangtise ... | head` 遇 stdout `EPIPE` 时安静退出；只有首个参数是 `--version` / `-V` 时才触发版本快捷路径
-- Endpoint registry 的 `key` 改为由记录键自动派生，减少映射漂移；新增真实 CLI 选项到请求体的 stub 测试；测试 272 → 323
-
-### v0.21.0 — 2026-06-29
-
-**行为变更（注意）**
-- ⚠️ `vault wechat-chatroom-list` 省略 `--size` 现在**拉全量**（此前默认只返回 20 条）。该接口不返回 `total`，CLI 改为串行翻页（翻到不满页为止，单页上限 50）；传 `--size N` 仍只取前 N 条。依赖"默认 20 条"的脚本会拿到全部群。
-
-**修复**
-- `quote day-kline --security all` 等大结果集用默认 `table` 格式输出时不再因 `Math.max(...大数组)` 撑爆调用栈崩溃（`RangeError`）；`renderTable` 改用 reduce 计算列宽
-- CSV 导出：含回车符 `\r` 的字段现在正确加引号（RFC 4180）；`table` / `markdown` 的多行字段折叠换行，保持表格对齐
-- 下载文件名剥离控制字符 / NUL，避免 `fs.writeFile` 报错
-
-**修复（安全）**
-- token 缓存文件（`~/.config/gangtise/token.json`）改为临时文件 + 原子 `rename` 写入：从第一字节即 `0600`，消除"旧文件宽松权限残留"与"崩溃截断"两个隐患
-
-**内部 / 工程**
-- 依赖 `vitest` 升级到 3.2.6（修复 dev-only 安全告警）；新增 `npm run typecheck`；测试 257 → 272
-
-### v0.20.0 — 2026-06-26
-
-**新增接口**
-- `insight announcement-us list` / `download` — 美股公司公告列表与下载（`--security TSLA.O`、`--category`〔分类用 `reference constant-list --category usShareAnnouncementCategory`，美股独立的 `103980xxx` 段〕、`--search-type`、`--rank-type`、下载 `--file-type 1` 原始 PDF / `2` Markdown）；自动翻页，单页上限 50
-- `ai stock-summary` — 个股看点（精炼投研总结）：`--security` 传具体代码（A股/港股，可重复，单次最多 6000）或市场关键词 `aShares` / `hkStocks` 拉全市场；无看点的证券不返回、不扣分
-- `fundamental income-statement-us` / `balance-sheet-us` / `cash-flow-us` — 美股三大财务报表（参数同其他财报：`--security-code` / `--period` / `--report-type` / `--fiscal-year` / `--field` 等）
-- `reference chiefs-search` — 首席分析师 ID 搜索（`--keyword` 按姓名/机构/团队匹配，`--top` 默认 10）；用于 `insight opinion list --chief` 的入参
-
-**变更**
-- `insight announcement-hk download` 新增 `--file-type`（`1` 原始（默认）/ `2` Markdown），此前无格式选项
-
-**行为变更（注意）**
-- ⚠️ `auth login` / `auth status` 默认脱敏 access token：`--format json` 输出里 `authorization` 与 `cache.accessToken` 显示为 `<redacted>`，仅保留过期时间 / 用户名 / 产品码 / uid 等非敏感字段。**依赖 `auth login` 原始 token 输出的脚本会拿到 `<redacted>`**，需改用 `auth login --show-token` 获取明文。
-
-**修复（安全）**
-- `auth status` / `auth login` token 脱敏：按凭证字段名模式匹配（`token`/`key`/`secret`/`password`/`credential`），覆盖 `apiKey`/`privateKey`/`refreshToken` 等任何可能携带的凭证字段
-- 自愈守卫：同时设 `GANGTISE_TOKEN` + AK/SK 时，注入 token 失效后重新登录不再被旧 token 短路，重试改用登录拿到的新 token
-
-**修复（数据正确性 / 健壮性）**
-- ⚠️ **CSV 负数不再被破坏**（影响所有 CSV 导出）：此前防公式注入会把负数（如跌幅 `-3.5`）加 `'` 前缀变成文本，Excel/pandas 无法参与计算；现仅对非有限数字的可疑串（`=`/`@`/`-1+cmd` 等）加前缀，合法数字原样输出
-- 自动翻页改为 fail-soft：某页遇不可重试错误（限流 `903301` 等）不再丢弃已取的全部数据，返回已取页 + `partial` / `failedPages` 标记，并在首错后停止继续请求（避免撞限流多烧配额）
-- 下载文件名 fallback（服务端 `Content-Disposition`）补清洗：含 `/`、`:` 等字符的文件名不再写到意外路径
-- `ai stock-summary` / `ai knowledge-batch` 缺 `--security` / `--query` 时本地报错，不再发空请求（stock-summary 借此避免被后台当全市场误扣积分）
-- `ai hot-topic` `--no-with-related-securities` / `--no-with-close-reading` 改为显式发 `false`（语义更明确，不依赖"字段缺失=排除"的隐含约定）
-
-**修复（indicator 适配 EDE 后台新结构）**
-- `indicator cross-section` / `time-series` 适配后台改版的返回结构（字段名加 `List` 后缀 `securityCodeList/indicatorCodeList/…`、截面 `values` 改二维 `[指标][证券]`）：此前后台改结构后 CLI 拍平失配、退化成原始矩阵，现恢复 `{date, security, name, 指标:值}` 宽表。配合后台同步变化——无数据从 `999999` 报错改为返回 `null`（截面不再 500、不丢行），缺必填参数从笼统 `410106` 改为直接指明缺哪个参数
-
 > 更早版本及完整更新历史见 [CHANGELOG.md](CHANGELOG.md)。
 
 ## 首次安装
@@ -242,9 +205,12 @@ cp -r gangtise-openapi ~/.hermes/skills/gangtise-openapi
 | | `foreign-opinion list` | 外资机构观点 |
 | | `independent-opinion list` / `download` | 外资独立分析师观点（含原文/翻译HTML下载） |
 | | `official-account list` / `download` | 产业公众号资讯（含 txt/HTML 下载） |
+| | `qa list` | 投资者问答 QA（互动平台/电话会议/调研纪要，按证券） |
+| | `report-image list` / `download` | 研报图表搜索（按关键词，含原图 JPEG 下载） |
 | **Reference** | `securities-search` | GTS Code 搜索（按名称/代码/拼音匹配） |
 | | `chiefs-search` | 首席分析师 ID 搜索（按姓名/机构/团队匹配） |
 | | `institution-search` | 机构 ID 搜索（内资券商/外资/牵头/观点机构，按名称匹配） |
+| | `official-account-search` | 公众号 ID 搜索（按公众号名/机构/分类匹配，返回 accountId） |
 | | `constant-category` | 常量分类列表（含各分类适用的接口与参数） |
 | | `constant-list` | 按分类导出常量值全量列表（行业/城市/公告分类/区域等） |
 | | `concept-search` | 题材 ID 搜索（名称/拼音/分组名匹配） |
@@ -336,9 +302,9 @@ gangtise ai knowledge-batch --query 比亚迪 --query 最近热门概念
 - **HTTP keep-alive**：所有请求复用同一个 `undici.Agent`（连接池 16），避免重复 TLS 握手。
 - **流式下载**：指定 `--output` 时，二进制响应（PDF 等）直接 `pipeline` 到磁盘，不经过内存缓冲；50MB PDF 内存占用近乎为零。
 - **流式输出**：`jsonl`/`csv` 格式且 `--output` 指定时，超过 1000 行自动切换为逐行写盘，避免一次性构建百 MB 字符串。
-- **自动重试**：5xx / 429 / `ECONNRESET` / `ETIMEDOUT` / `ENOTFOUND` / `EAI_AGAIN` / `UND_ERR_*`（undici 超时类）/ `999999` 系统错误自动指数退避重试 2 次。
+- **自动重试**：5xx / 429 / `ECONNREFUSED` / `ECONNRESET` / `ETIMEDOUT` / `ENOTFOUND` / `EAI_AGAIN` / `UND_ERR_*`（undici 连接/超时类）/ `999999` 系统错误自动指数退避重试 2 次。**贵档生成类端点例外**（one-pager 等 13 个）：5xx/超时不重放——按次计费不幂等，重放即重复扣分；仅连接失败、429 与 token 自愈重试。
 - **Token 自愈**：调用返回 8000014/8000015 时自动强制刷新 Token 并重试一次。
-- **K线自动分片**：`quote day-kline --security all` 等全市场查询自动按日期切分（A股 1 天/片、美股 1 天/片、HK 2 天/片、指数 30 天/片），并发执行后合并结果。分片时如果用户未传 `--limit`，自动注入 `limit: 10000`（API 上限）避免默认 6000 截断。
+- **K线/资金流向自动分片**：`quote day-kline --security all`、`quote fund-flow --security aShares` 等全市场查询自动按日期切分（A股 K线/资金流向 1 天/片、美股 1 天/片、HK 2 天/片、指数 30 天/片），并发执行后合并结果；按日分片自动跳过周六日。分片时如果用户未传 `--limit`，自动注入 `limit: 10000`（API 上限）避免默认 6000 截断。
 - **Token 内存缓存**：Token 在进程内存中缓存，避免每次请求读盘。
 - **`--verbose`**：打印每个请求的方法、路径、状态码、耗时和响应大小到 stderr，方便定位慢查询。
 
@@ -359,6 +325,7 @@ gangtise ai knowledge-batch --query 比亚迪 --query 最近热门概念
 - `insight foreign-opinion list`
 - `insight independent-opinion list`
 - `insight official-account list`
+- `insight qa list`
 - `ai security-clue`
 - `vault drive-list`
 - `vault record-list`
@@ -447,6 +414,12 @@ gangtise insight independent-opinion download --independent-opinion-id 207051900
 gangtise insight official-account list --keyword 泡泡玛特 --rank-type 2 --size 20 --format json
 gangtise insight official-account download --article-id 7286248 --file-type 2
 
+# 投资者问答 QA（按证券；--source/--question-category/--answer-important 精筛，自动翻页）
+gangtise insight qa list --security-code 601012.SH --source interactive --answer-important 1 --size 20 --format json
+# 研报图表：按关键词搜图拿 chunkId，再下原图（JPEG）
+gangtise insight report-image list --keyword AI --top 5 --format json
+gangtise insight report-image download --chunk-id image_10_384655917758685184_8 --output ./ai-chart.jpg
+
 # 纪要下载（会议平台来源可选 HTML 格式）
 gangtise insight summary download --summary-id 4906813 --file-type 2
 ```
@@ -463,8 +436,10 @@ gangtise reference securities-search --keyword "银行" --category stock --categ
 # 首席分析师 ID 搜索（按姓名/机构/团队；拿 chiefId 供 insight opinion list --chief 使用）
 gangtise reference chiefs-search --keyword 东吴证券 --top 3 --format json
 gangtise reference chiefs-search --keyword 芦哲 --format json
-# 机构 ID 搜索（--category: domesticBroker/foreignInstitution/leadInstitution/opinionInstitution）
+# 机构 ID 搜索（--category: domesticBroker/foreignInstitution/leadInstitution/opinionInstitution/foreignOpinionInstitution）
 gangtise reference institution-search --keyword 招商证券 --category domesticBroker --top 3 --format json
+# 公众号 ID 搜索（按名称/机构/分类；拿 accountId 供 insight official-account list --account-id）
+gangtise reference official-account-search --keyword 中信证券 --top 3 --format json
 
 # 常量查询：先看分类，再按分类导出全量常量值
 gangtise reference constant-category --format json

package/dist/src/cli.js

@@ -2,7 +2,7 @@
 import { Command, Option } from "commander";
 import { checkAsyncContent, pollAsyncContent, POLL_MAX_ATTEMPTS } from "./core/asyncContent.js";
 import { readTokenCache, redactTokenCache } from "./core/auth.js";
-import { collectKeyValue, collectList, collectNumberList, localDateString, maybeArray, parseFrom, parseNumberOption, parseOptionalNumberOption, parseSize, parseTimestamp13 } from "./core/args.js";
+import { collectKeyValue, collectList, collectNumberList, localDateString, maybeArray, parseChoiceList, parseFrom, parseNumberOption, parseOptionalNumberOption, parseSize, parseTimestamp13 } from "./core/args.js";
 import { buildIndicatorCrossSectionBody, buildIndicatorTimeSeriesBody, buildQuoteKlineBody, buildStockPoolStocksBody, buildWechatChatroomListBody, buildWechatMessageListBody } from "./core/commandBodies.js";
 import { flattenCrossSection, flattenTimeSeries, unwrapIndicatorData } from "./core/indicatorMatrix.js";
 import { callKlineWithSharding, isAllMarket, isFullMarket } from "./core/quoteSharding.js";
@@ -169,6 +169,8 @@ const announcementUs = new Command("announcement-us");
 const foreignOpinion = new Command("foreign-opinion");
 const independentOpinion = new Command("independent-opinion");
 const officialAccount = new Command("official-account");
+const qa = new Command("qa");
+const reportImage = new Command("report-image");
 addTimeFilters(opinion.command("list").option("--rank-type <number>", "Rank type", "1").option("--research-area <id>", "Research area ID", collectList, []).option("--chief <id>", "Chief ID", collectList, []).option("--security <code>", "Security code", collectList, []).option("--broker <id>", "Broker ID", collectList, []).option("--industry <id>", "Industry ID", collectList, []).option("--concept <id>", "Concept ID", collectList, []).option("--llm-tag <tag>", "Semantic tag", collectList, []).option("--source <source>", "Source", collectList, []).option("--format <format>", "Output format", "table").option("--output <path>", "Output path")).action((options) => emit(options, (client) => client.call("insight.opinion.list", {
     from: parseFrom(options.from), size: parseSize(options.size), startTime: options.startTime, endTime: options.endTime,
     rankType: parseNumberOption(options.rankType, "--rank-type", { integer: true, min: 1 }), keyword: options.keyword, researchAreaList: maybeArray(options.researchArea), chiefList: maybeArray(options.chief),
@@ -310,6 +312,19 @@ addTimeFilters(officialAccount.command("list").option("--search-type <number>",
     categoryList: maybeArray(options.category), industryList: maybeArray(options.industry),
 }), { endpointKey: "insight.official-account.list", idField: "articleId" }));
 addDownloadCommand(officialAccount, { endpointKey: "insight.official-account.download", idOption: "--article-id", idField: "articleId", fallbackPrefix: "official-account", fileType: { description: "File type: 1=txt(default) 2=HTML", default: "1" }, titleListEndpoint: "insight.official-account.list" });
+// QA request keys are BARE (source/questionCategory/answerImportant), not the *List
+// convention — the body below mirrors the spec exactly. Datetimes pass through as strings.
+qa.command("list").requiredOption("--security-code <code>", "Security code, e.g. 601012.SH").option("--from <number>", "Starting offset", "0").option("--size <number>", "Total rows to return; omit to fetch all (max page 500)").option("--start-time <datetime>", "Start time (yyyy-MM-dd or yyyy-MM-dd HH:mm:ss)").option("--end-time <datetime>", "End time (yyyy-MM-dd or yyyy-MM-dd HH:mm:ss)").option("--source <type>", "Source: conference/interactive/survey (repeat)", collectList, []).option("--question-category <name>", "Question category (repeat): productAndBusiness/capacityAndProjects/ordersAndCustomers/financialData/materialEvents/capitalOperations/shareholdersAndDividends/corporateGovernance/marketAndValuation/macroAndIndustry/risksAndOthers", collectList, []).option("--answer-important <flag>", "Answer involves key info: 1=yes 0=no (repeat; omit for all)", collectNumberList, []).option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("insight.qa.list", {
+    from: parseFrom(options.from), size: parseSize(options.size),
+    securityCode: options.securityCode, startTime: options.startTime, endTime: options.endTime,
+    source: maybeArray(options.source), questionCategory: maybeArray(options.questionCategory),
+    answerImportant: options.answerImportant.length ? options.answerImportant : undefined,
+})));
+reportImage.command("list").requiredOption("--keyword <text>", "Search keyword, e.g. 'AI' '新能源汽车'").option("--top <number>", "Max results (default: 10, max: 20)", "10").option("--source-id <id>", "Report source ID, to filter to one report (from a report list or knowledge base)").option("--start-time <datetime>", "Start time (yyyy-MM-dd HH:mm:ss; yyyy-MM-dd auto-completed)").option("--end-time <datetime>", "End time (yyyy-MM-dd HH:mm:ss; yyyy-MM-dd auto-completed)").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("insight.report-image.list", {
+    keyword: options.keyword, top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 20 }),
+    sourceId: options.sourceId, startTime: options.startTime, endTime: options.endTime,
+})));
+addDownloadCommand(reportImage, { endpointKey: "insight.report-image.download", idOption: "--chunk-id", idField: "chunkId", fallbackPrefix: "report-image" });
 insight.addCommand(opinion);
 insight.addCommand(summary);
 insight.addCommand(roadshow);
@@ -324,6 +339,8 @@ insight.addCommand(announcementUs);
 insight.addCommand(foreignOpinion);
 insight.addCommand(independentOpinion);
 insight.addCommand(officialAccount);
+insight.addCommand(qa);
+insight.addCommand(reportImage);
 program.addCommand(insight);
 const quote = new Command("quote").description("Quote APIs");
 const addKlineCommand = (name, endpointKey, securityHelp, shardDays) => quote.command(name)
@@ -436,10 +453,10 @@ fundamental.command("earning-forecast").requiredOption("--security-code <code>")
 }));
 program.addCommand(fundamental);
 const ai = new Command("ai").description("AI APIs");
-ai.command("knowledge-batch").option("--query <text>", "Query", collectList, []).option("--top <number>", "Top", "10").option("--resource-type <number>", "Resource type", collectNumberList, []).option("--knowledge-name <name>", "Knowledge name", collectList, []).option("--start-time <ms>").option("--end-time <ms>").option("--format <format>", "Output format", "json").option("--output <path>").action((options) => {
+ai.command("knowledge-batch").option("--query <text>", "Query", collectList, []).option("--top <number>", "Max results (default: 10, max: 20)", "10").option("--resource-type <number>", "Resource type", collectNumberList, []).option("--knowledge-name <name>", "Knowledge name", collectList, []).option("--start-time <ms>").option("--end-time <ms>").option("--format <format>", "Output format", "json").option("--output <path>").action((options) => {
     if (!options.query.length)
         throw new ValidationError("--query is required: pass at least one --query");
-    return emit(options, (client) => client.call("ai.knowledge-batch", { queries: options.query, top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }), resourceTypes: options.resourceType.length ? options.resourceType : undefined, knowledgeNames: maybeArray(options.knowledgeName), startTime: parseOptionalNumberOption(options.startTime, "--start-time", { integer: true, min: 0 }), endTime: parseOptionalNumberOption(options.endTime, "--end-time", { integer: true, min: 0 }) }));
+    return emit(options, (client) => client.call("ai.knowledge-batch", { queries: options.query, top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 20 }), resourceTypes: options.resourceType.length ? options.resourceType : undefined, knowledgeNames: maybeArray(options.knowledgeName), startTime: parseOptionalNumberOption(options.startTime, "--start-time", { integer: true, min: 0 }), endTime: parseOptionalNumberOption(options.endTime, "--end-time", { integer: true, min: 0 }) }));
 });
 ai.command("knowledge-resource-download").requiredOption("--resource-type <number>").requiredOption("--source-id <id>").option("--output <path>").action((options) => withClient(async (client) => {
     await runDownload(client, "ai.knowledge-resource.download", { resourceType: parseNumberOption(options.resourceType, "--resource-type", { integer: true, min: 0 }), sourceId: options.sourceId }, {
@@ -538,28 +555,34 @@ ai.command("stock-summary").description("Stock highlights: refined research summ
 const reference = new Command("reference").description("Reference data APIs");
 reference.command("securities-search").requiredOption("--keyword <text>", "Search keyword (name/code/pinyin/English)").option("--category <type>", "Category: stock/dr/index/fund", collectList, []).option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.securities-search", {
     keyword: options.keyword,
-    category: options.category.length ? options.category : undefined,
-    top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }),
+    category: parseChoiceList(options.category, "--category", ["stock", "dr", "index", "fund"]),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
 })));
 reference.command("constant-category").description("List constant categories and which API params accept them").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.constant-category")));
 reference.command("constant-list").requiredOption("--category <code>", "Category code from 'reference constant-category' (e.g. citicIndustry/swIndustry/regionCategory)").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.constant-list", { category: options.category })));
 reference.command("concept-search").requiredOption("--keyword <text>", "Search keyword (name/pinyin/group name)").option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.concept-search", {
     keyword: options.keyword,
-    top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
 })));
 reference.command("sector-search").option("--keyword <text>", "Search keyword (name/pinyin)").option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.sector-search", {
     keyword: options.keyword,
-    top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
 })));
 reference.command("sector-constituents").requiredOption("--sector-id <id>", "Sector ID from 'reference sector-search'").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.sector-constituents", { sectorId: options.sectorId })));
 reference.command("chiefs-search").requiredOption("--keyword <text>", "Search keyword (chief name / institution / team)").option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.chiefs-search", {
     keyword: options.keyword,
-    top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
 })));
 reference.command("institution-search").requiredOption("--keyword <text>", "Search keyword (institution name / abbreviation)").option("--category <name>", "Category: domesticBroker/foreignInstitution/leadInstitution/opinionInstitution/foreignOpinionInstitution (repeat); omit for all", collectList, []).option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.institution-search", {
     keyword: options.keyword,
-    categoryList: maybeArray(options.category),
-    top: parseNumberOption(options.top, "--top", { integer: true, min: 1 }),
+    categoryList: parseChoiceList(options.category, "--category", ["domesticBroker", "foreignInstitution", "leadInstitution", "opinionInstitution", "foreignOpinionInstitution"]),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
+})));
+// Note: request key is BARE `category` here (spec), unlike institution-search's `categoryList`.
+reference.command("official-account-search").requiredOption("--keyword <text>", "Search keyword (account name / institution / keyword, e.g. 东吴证券)").option("--category <name>", "Category: listedCompany/broker/government/media (repeat); omit for all incl. uncategorized", collectList, []).option("--top <number>", "Max results (default: 10, max: 10)", "10").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("reference.official-account-search", {
+    keyword: options.keyword,
+    category: parseChoiceList(options.category, "--category", ["listedCompany", "broker", "government", "media"]),
+    top: parseNumberOption(options.top, "--top", { integer: true, min: 1, max: 10 }),
 })));
 program.addCommand(reference);
 const vault = new Command("vault").description("Vault APIs");
@@ -578,7 +601,7 @@ program.addCommand(ai);
 const alternative = new Command("alternative").description("Alternative data APIs");
 alternative.command("edb-search").requiredOption("--keyword <text>", "Search keyword (e.g. '空调')").option("--limit <number>", "Max results (default: 100, max: 200)", "100").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => emit(options, (client) => client.call("alternative.edb-search", {
     keyword: options.keyword,
-    limit: parseNumberOption(options.limit, "--limit", { integer: true, min: 1 }),
+    limit: parseNumberOption(options.limit, "--limit", { integer: true, min: 1, max: 200 }),
 })));
 alternative.command("edb-data").option("--indicator-id <id>", "Indicator ID (repeat, max 10)", collectList, []).requiredOption("--start-date <date>", "Start date (yyyy-MM-dd)").requiredOption("--end-date <date>", "End date (yyyy-MM-dd)").option("--format <format>", "Output format", "table").option("--output <path>").action((options) => withClient(async (client) => {
     const format = parseOutputFormat(options.format);
@@ -605,7 +628,7 @@ indicator.command("search").requiredOption("--keyword <text>", "Search keyword,
     const format = parseOutputFormat(options.format);
     const raw = await client.call("indicator.search", {
         keyword: options.keyword,
-        limit: parseNumberOption(options.limit, "--limit", { integer: true, min: 1 }),
+        limit: parseNumberOption(options.limit, "--limit", { integer: true, min: 1, max: 100 }),
     });
     await printData(unwrapIndicatorData(raw), format, options.output);
 }));

package/dist/src/core/args.js

@@ -62,6 +62,17 @@ export function collectKeyValue(value, previous = {}) {
 export function maybeArray(value) {
     return value.length > 0 ? value : undefined;
 }
+// Whitelist for enum-valued repeatable options. Only used where the server was
+// probed NOT to reject bad values (it silently ignores the filter or returns
+// empty instead) — endpoints that answer 100003 keep server-side validation.
+export function parseChoiceList(values, optionName, allowed) {
+    for (const value of values) {
+        if (!allowed.includes(value)) {
+            throw new ValidationError(`Invalid ${optionName}: "${value}" is not one of ${allowed.join("/")}`);
+        }
+    }
+    return maybeArray(values);
+}
 export function toTimestamp13(value) {
     if (value === undefined)
         return undefined;

package/dist/src/core/client.js

@@ -285,13 +285,30 @@ export class GangtiseClient {
             total,
             list: requestedSize === undefined ? collected : collected.slice(0, requestedSize),
         };
-        if (failedPages.length > 0) {
+        // Unified completeness backstop. Whatever the cause — a failed/shape-broken page,
+        // a short later page (server page cap < maxPageSize), the MAX_PAGES cap, or `total`
+        // drifting mid-fetch — the result is partial. A short row count or a total drift each
+        // force it; so does any failedPages entry on its own — an over-returning sibling page
+        // can lift the row count back to target and mask the hole (short would read false), yet
+        // the failedPages branch below still writes "results are partial" to stderr, so the flag
+        // must agree. printData maps partial → exit 3 so a script can't read a truncated export
+        // as complete. The cap and drift branches above already warned on stderr; failedPages
+        // warns below.
+        const short = collected.length < target;
+        if (short || totalDrift || failedPages.length > 0)
             out.partial = true;
+        if (failedPages.length > 0) {
             out.failedPages = failedPages.map((p) => ({ from: p.from, size: p.size }));
             const detail = firstError instanceof Error ? `: ${firstError.message}` : "";
             const skippedHint = aborted ? " A page hit a non-retryable error (e.g. rate limit); remaining pages were skipped." : "";
             process.stderr.write(`[gangtise] warning: ${failedPages.length}/${pageRequests.length} pages not fetched${detail}; results are partial — got ${collected.length}/${total} rows (see failedPages).${skippedHint}\n`);
         }
+        else if (short && !truncatedByPageCap && !totalDrift) {
+            // A short later page with no failure, cap, or drift to explain it: the server
+            // simply delivered fewer rows than `total` promised. Warn so an interactive run
+            // sees why the result is partial (the other causes each warn on their own path).
+            process.stderr.write(`[gangtise] warning: server returned ${collected.length} of ${total} rows (a later page came back short); results may be incomplete\n`);
+        }
         return out;
     }
     async login() {
@@ -337,9 +354,22 @@ export class GangtiseClient {
             // response reads as text directly (and keeps existing behavior on that path).
             const encoding = response.headers['content-encoding'];
             const gzipped = (Array.isArray(encoding) ? encoding[0] : encoding)?.toLowerCase().trim() === 'gzip';
-            const text = gzipped
-                ? decodeResponseBody(new Uint8Array(await response.body.arrayBuffer()), encoding)
-                : await response.body.text();
+            let text;
+            if (gzipped) {
+                // A proxy/middlebox can declare gzip and deliver garbage — surface that as
+                // an ApiError with request context instead of a bare zlib Z_DATA_ERROR.
+                const bytes = new Uint8Array(await response.body.arrayBuffer());
+                try {
+                    text = decodeResponseBody(bytes, encoding);
+                }
+                catch (error) {
+                    const detail = error instanceof Error ? error.message : String(error);
+                    throw new ApiError(`Failed to decode gzip response for ${endpoint.method} ${endpoint.path}: ${detail}`, undefined, response.statusCode);
+                }
+            }
+            else {
+                text = await response.body.text();
+            }
             logTiming(`${endpoint.method} ${endpoint.path}`, Date.now() - startedAt, `${response.statusCode}, ${text.length}B`);
             // Parse Retry-After once so every error path below (JSON parse failure AND the
             // envelope/HTTP-error throw) carries it — a non-JSON 429/503 must still honor
@@ -368,6 +398,7 @@ export class GangtiseClient {
                 throw error;
             }
         }, {
+            policy: endpoint.retry === "no-replay" ? "no-replay" : "default",
             onRetry: (attempt, error, delay) => {
                 if (!isVerbose())
                     return;
@@ -417,6 +448,13 @@ export class GangtiseClient {
                     dispatcher,
                 });
             }
+            // The loop above can exit with a 3xx still in hand (hop limit exceeded, or a
+            // redirect without Location) — that response must never be treated as file
+            // content: its HTML placeholder body would be saved as the "downloaded file".
+            if (response.statusCode >= 300 && response.statusCode < 400) {
+                await response.body.text().catch(() => { });
+                throw new ApiError(`Download failed: unresolved redirect (HTTP ${response.statusCode})`, undefined, response.statusCode);
+            }
             const contentType = Array.isArray(response.headers['content-type']) ? response.headers['content-type'][0] : response.headers['content-type'];
             // From the final (post-redirect) response, so a rate-limited download honors
             // Retry-After too — every error branch below passes it into the ApiError.
@@ -482,14 +520,17 @@ export class GangtiseClient {
             // Stream directly to disk when caller already knows the destination
             if (options?.streamTo) {
                 await fs.mkdir(path.dirname(options.streamTo), { recursive: true });
+                // Stream into a .part sibling and rename over the target only on success:
+                // writing to the target directly would truncate an existing file on the
+                // FIRST byte and delete it on failure — a failed re-download (or each
+                // withRetry attempt) must never destroy the user's previous good file.
+                const partPath = `${options.streamTo}.part`;
                 try {
-                    await pipeline(response.body, createWriteStream(options.streamTo));
+                    await pipeline(response.body, createWriteStream(partPath));
+                    await fs.rename(partPath, options.streamTo);
                 }
                 catch (error) {
-                    // A mid-stream failure leaves a truncated file on disk; remove it so a
-                    // failed download never looks like a complete one. withRetry may still
-                    // replay the request (the next attempt re-creates the file).
-                    await fs.unlink(options.streamTo).catch(() => { });
+                    await fs.unlink(partPath).catch(() => { });
                     throw error;
                 }
                 logTiming(`GET ${endpoint.path} (stream)`, Date.now() - startedAt, `${response.statusCode}`);

package/dist/src/core/download.js

@@ -26,7 +26,9 @@ function truncateFilename(name, maxBytes = 200) {
 /** Pick a non-existing path by suffixing -1, -2, … before the extension, so batch
  * downloads whose titles collide ("2025年第一季度报告" from several companies) don't
  * silently overwrite each other. Only auto-derived names go through this — an
- * explicit --output path keeps plain overwrite semantics. */
+ * explicit --output path keeps plain overwrite semantics. Throws instead of
+ * falling back to the original path once the suffixes run out: returning `p`
+ * there would silently overwrite the very first file. */
 export async function uniquePath(p) {
     const exists = (f) => fs.access(f).then(() => true, () => false);
     if (!(await exists(p)))
@@ -38,7 +40,7 @@ export async function uniquePath(p) {
         if (!(await exists(candidate)))
             return candidate;
     }
-    return p;
+    throw new DownloadError(`Refusing to overwrite: 100 files already share the name "${p}" — pass --output or clean up the directory`);
 }
 const MIME_EXT = {
     "application/pdf": ".pdf",
@@ -106,21 +108,70 @@ export async function resolveTitle(client, result, listEndpoint, idField, idValu
 }
 async function downloadUrlTo(url, outputPath) {
     const { createWriteStream } = await import("node:fs");
-    const { Readable } = await import("node:stream");
     const { pipeline } = await import("node:stream/promises");
     const { dirname } = await import("node:path");
-    const response = await fetch(url);
-    if (!response.ok || !response.body) {
-        throw new DownloadError(`Failed to fetch download URL (HTTP ${response.status})`);
-    }
+    const { request } = await import("undici");
+    const { getDispatcher, logTiming, withRetry } = await import("./transport.js");
+    const { loadConfig } = await import("./config.js");
+    // Through the transport layer instead of a bare global fetch: the configured
+    // timeout applies (a slow-drip CDN can no longer hang the CLI indefinitely),
+    // network-level failures retry, and --verbose logs the request.
+    const timeoutMs = loadConfig().timeoutMs;
     await fs.mkdir(dirname(outputPath), { recursive: true });
-    try {
-        await pipeline(Readable.fromWeb(response.body), createWriteStream(outputPath));
-    }
-    catch (error) {
-        await fs.unlink(outputPath).catch(() => { });
-        throw error;
-    }
+    // Signed URLs carry credentials in the query string — verbose logs must strip
+    // search/hash so signatures never land in terminal/CI logs.
+    const redactUrl = (u) => {
+        try {
+            const parsed = new URL(u);
+            return parsed.origin + parsed.pathname;
+        }
+        catch {
+            return "signed-url";
+        }
+    };
+    // .part + rename so a failed follow-download never destroys an existing file.
+    const partPath = `${outputPath}.part`;
+    await withRetry(async () => {
+        const startedAt = Date.now();
+        const requestOptions = {
+            method: "GET",
+            headersTimeout: timeoutMs,
+            bodyTimeout: timeoutMs,
+            dispatcher: getDispatcher(),
+        };
+        let currentUrl = url;
+        let response = await request(currentUrl, requestOptions);
+        // undici does not follow redirects (the old global fetch did): a signed URL
+        // may 302 to the real CDN object. Follow a bounded number of hops; no
+        // Authorization header is involved on this path, so nothing leaks cross-origin.
+        for (let hops = 0; hops < 3 && response.statusCode >= 300 && response.statusCode < 400; hops++) {
+            const locationHeader = response.headers.location;
+            const location = Array.isArray(locationHeader) ? locationHeader[0] : locationHeader;
+            if (!location)
+                break;
+            await response.body.text().catch(() => { });
+            currentUrl = new URL(location, currentUrl).toString();
+            response = await request(currentUrl, requestOptions);
+        }
+        if (response.statusCode >= 300 && response.statusCode < 400) {
+            await response.body.text().catch(() => { });
+            throw new DownloadError(`Failed to fetch download URL: unresolved redirect (HTTP ${response.statusCode})`);
+        }
+        if (response.statusCode >= 400) {
+            await response.body.text().catch(() => { });
+            // DownloadError (not retried): signed URLs expire — replaying a 403 is useless.
+            throw new DownloadError(`Failed to fetch download URL (HTTP ${response.statusCode})`);
+        }
+        try {
+            await pipeline(response.body, createWriteStream(partPath));
+        }
+        catch (error) {
+            await fs.unlink(partPath).catch(() => { });
+            throw error;
+        }
+        logTiming(`GET ${redactUrl(currentUrl)}`, Date.now() - startedAt, String(response.statusCode));
+    });
+    await fs.rename(partPath, outputPath);
 }
 export async function saveDownloadResult(result, fallbackName, output) {
     if (!(result && typeof result === "object")) {

package/dist/src/core/endpoints.js

@@ -173,6 +173,26 @@ const ENDPOINT_DEFS = {
         kind: "download",
         description: "Download WeChat official account article (txt/HTML)",
     },
+    "insight.qa.list": {
+        method: "POST",
+        // The literal '&' is the vendor's path segment (Q&A-data), not a query separator.
+        path: "/application/open-insight/Q&A-data/getList",
+        kind: "json",
+        description: "List investor Q&A (conference/interactive/survey) for a security",
+        pagination: { enabled: true, maxPageSize: 500 },
+    },
+    "insight.report-image.list": {
+        method: "POST",
+        path: "/application/open-insight/report-image/getList",
+        kind: "json",
+        description: "Search research report images by keyword (returns chunkId + metadata)",
+    },
+    "insight.report-image.download": {
+        method: "GET",
+        path: "/application/open-insight/report-image/download/file",
+        kind: "download",
+        description: "Download a research report image by chunkId",
+    },
     // ─── reference ───
     "reference.securities-search": {
         method: "POST",
@@ -192,6 +212,12 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Search institution IDs by keyword (domestic broker / foreign / lead / opinion institution)",
     },
+    "reference.official-account-search": {
+        method: "POST",
+        path: "/application/open-reference/officialAccount/search",
+        kind: "json",
+        description: "Search official account (WeChat public account) IDs by name / institution / category",
+    },
     "reference.constant-category": {
         method: "GET",
         path: "/application/open-reference/constants/category",
@@ -368,6 +394,7 @@ const ENDPOINT_DEFS = {
         path: "/application/open-data/ai/search/knowledge/batch",
         kind: "json",
         description: "Batch knowledge search",
+        retry: "no-replay",
     },
     "ai.knowledge-resource.download": {
         method: "GET",
@@ -388,6 +415,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Generate one pager",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.investment-logic": {
         method: "POST",
@@ -395,6 +423,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Generate investment logic",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.peer-comparison": {
         method: "POST",
@@ -402,12 +431,14 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Generate peer comparison",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.earnings-review.get-id": {
         method: "POST",
         path: "/application/open-ai/agent/earnings-review-getid",
         kind: "json",
         description: "Get earnings review ID",
+        retry: "no-replay",
     },
     "ai.earnings-review.get-content": {
         method: "POST",
@@ -421,6 +452,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Get theme tracking daily report",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.research-outline": {
         method: "POST",
@@ -428,6 +460,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Get company research outline",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.hot-topic": {
         method: "POST",
@@ -435,6 +468,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "List hot topic reports",
         pagination: { enabled: true, maxPageSize: 20 },
+        retry: "no-replay",
     },
     "ai.management-discuss-announcement": {
         method: "POST",
@@ -442,6 +476,7 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Management discussion from financial reports (half-year/annual)",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.management-discuss-earnings-call": {
         method: "POST",
@@ -449,12 +484,14 @@ const ENDPOINT_DEFS = {
         kind: "json",
         description: "Management discussion from earnings calls",
         timeoutMs: 120_000,
+        retry: "no-replay",
     },
     "ai.viewpoint-debate.get-id": {
         method: "POST",
         path: "/application/open-ai/agent/viewpoint-debate-getid",
         kind: "json",
         description: "Get viewpoint debate ID",
+        retry: "no-replay",
     },
     "ai.viewpoint-debate.get-content": {
         method: "POST",
@@ -547,12 +584,14 @@ const ENDPOINT_DEFS = {
         path: "/application/open-alternative/concept/info",
         kind: "json",
         description: "Query latest concept (theme index) profile by conceptId",
+        retry: "no-replay",
     },
     "alternative.concept-securities": {
         method: "POST",
         path: "/application/open-alternative/concept/securities",
         kind: "json",
         description: "Query concept (theme index) constituent securities, grouped",
+        retry: "no-replay",
     },
     // ─── indicator (EDE: security-level data indicators) ───
     "indicator.search": {

package/dist/src/core/errors.js

@@ -16,6 +16,7 @@ const ERROR_HINTS = {
     "999995": "当前账号积分不足。",
     "900002": "请求缺少 uid。",
     "900001": "请求参数为空或缺少必填项。",
+    "100003": "参数值非法——服务端不会指明是哪个参数，多为枚举参数拼写或取值超范围（如 --source / --question-category / --answer-important），对照命令 --help 列出的合法值检查。",
     "0000001008": "Token 已失效（多为他处登录挤掉本会话）；有 AK/SK 时 CLI 会自动重新登录重试一次，否则请重新登录。",
     "8000014": "GANGTISE_ACCESS_KEY 错误。",
     "8000015": "GANGTISE_SECRET_KEY 错误。",

package/dist/src/core/output.js

@@ -147,7 +147,10 @@ export async function streamOutputToFile(value, format, outputPath) {
     const { dirname } = await import("node:path");
     const { createWriteStream } = await import("node:fs");
     await fs.mkdir(dirname(outputPath), { recursive: true });
-    const stream = createWriteStream(outputPath, { encoding: "utf8" });
+    // Stream into a .part sibling and rename over the target only on success, so a
+    // failed re-export never destroys the previous good file.
+    const partPath = `${outputPath}.part`;
+    const stream = createWriteStream(partPath, { encoding: "utf8" });
     // A stream 'error' with no listener (EACCES on open, ENOSPC mid-write) crashes the
     // process before any write callback fires. Swallow the event here — the failure
     // still surfaces through the write/end callbacks below.
@@ -169,11 +172,20 @@ export async function streamOutputToFile(value, format, outputPath) {
         await new Promise((resolve, reject) => {
             stream.end((err) => err ? reject(err) : resolve());
         });
+        await fs.rename(partPath, outputPath);
     }
     catch (error) {
         // Mirror the download path: never leave a truncated file that looks complete.
+        // createWriteStream opens lazily — an early abort can reach unlink BEFORE the
+        // open() creates the file, so wait for 'close' (fd released or open aborted)
+        // or the .part would reappear right after being "removed".
         stream.destroy();
-        await fs.unlink(outputPath).catch(() => { });
+        // Already-closed happens when the failure was the rename (stream ended fine);
+        // then 'close' has fired and waiting for it again would hang forever.
+        if (!stream.closed) {
+            await new Promise((resolve) => stream.once("close", resolve));
+        }
+        await fs.unlink(partPath).catch(() => { });
         throw error;
     }
     return true;
@@ -222,9 +234,20 @@ export async function saveOutputIfNeeded(content, outputPath) {
     }
     const { dirname } = await import("node:path");
     await fs.mkdir(dirname(outputPath), { recursive: true });
-    if (typeof content === "string") {
-        await fs.writeFile(outputPath, content, "utf8");
-        return;
+    // Write to a .part sibling and rename over the target only on success, so a
+    // failed re-export/download never destroys the previous good file.
+    const partPath = `${outputPath}.part`;
+    try {
+        if (typeof content === "string") {
+            await fs.writeFile(partPath, content, "utf8");
+        }
+        else {
+            await fs.writeFile(partPath, content);
+        }
+        await fs.rename(partPath, outputPath);
+    }
+    catch (error) {
+        await fs.unlink(partPath).catch(() => { });
+        throw error;
     }
-    await fs.writeFile(outputPath, content);
 }

package/dist/src/core/quoteSharding.js

@@ -121,13 +121,31 @@ export async function callKlineWithSharding(client, endpointKey, body, config) {
     // every shard on the first rejection.
     const failedShards = [];
     let firstError = null;
+    let aborted = false;
     const results = await runWithConcurrency(shards, config.concurrency ?? PAGE_CONCURRENCY, async (shard) => {
+        // A prior shard hit a hard error (rate limit, no-perm, retries exhausted). Stop
+        // dispatching the rest rather than burning quota into the same failure; record them
+        // as failed so the merged result is flagged partial. Mirrors requestPaginated.
+        if (aborted) {
+            failedShards.push(shard);
+            return null;
+        }
         try {
-            return await client.call(endpointKey, { ...allMarketBody, startDate: shard.startDate, endDate: shard.endDate });
+            const res = await client.call(endpointKey, { ...allMarketBody, startDate: shard.startDate, endDate: shard.endDate });
+            // A shard that resolves but carries no `list` array is shape-broken (an error
+            // object, a truncated envelope) — its rows are missing. Treat it exactly like a
+            // thrown shard so the result is marked partial, not silently short. Unlike a hard
+            // error this doesn't abort the fan-out: one malformed shard isn't systemic.
+            if (!(res && typeof res === "object" && Array.isArray(res.list))) {
+                failedShards.push(shard);
+                return null;
+            }
+            return res;
         }
         catch (error) {
             if (!firstError)
                 firstError = error;
+            aborted = true;
             failedShards.push(shard);
             return null;
         }

package/dist/src/core/transport.js

@@ -65,13 +65,20 @@ export async function runWithConcurrency(items, concurrency, fn) {
 /** Fan-out width for pagination and kline shards — one env knob tunes both. */
 export const PAGE_CONCURRENCY = Number(process.env.GANGTISE_PAGE_CONCURRENCY ?? 5) || 5;
 const RETRYABLE_HTTP_STATUS = new Set([429, 500, 502, 503, 504]);
-const RETRYABLE_NETWORK_CODES = new Set(["ECONNRESET", "ETIMEDOUT", "ENOTFOUND", "EAI_AGAIN", "UND_ERR_SOCKET", "UND_ERR_HEADERS_TIMEOUT", "UND_ERR_BODY_TIMEOUT"]);
+const RETRYABLE_NETWORK_CODES = new Set(["ECONNREFUSED", "ECONNRESET", "ETIMEDOUT", "ENOTFOUND", "EAI_AGAIN", "UND_ERR_CONNECT_TIMEOUT", "UND_ERR_SOCKET", "UND_ERR_HEADERS_TIMEOUT", "UND_ERR_BODY_TIMEOUT"]);
 const RETRYABLE_API_CODES = new Set(["999999"]);
-function isRetryableError(error) {
+// Connect-phase / DNS failures: the request provably never reached the server, so a
+// replay cannot double-execute (or double-bill) anything even under "no-replay".
+const NO_REPLAY_NETWORK_CODES = new Set(["ECONNREFUSED", "ENOTFOUND", "EAI_AGAIN", "UND_ERR_CONNECT_TIMEOUT"]);
+function isRetryableError(error, policy) {
     if (error && typeof error === "object" && error.__retryable === true) {
         return true;
     }
     if (error instanceof ApiError) {
+        if (error.statusCode === 429)
+            return true;
+        if (policy === "no-replay")
+            return false;
         if (error.statusCode != null && RETRYABLE_HTTP_STATUS.has(error.statusCode))
             return true;
         if (error.code && RETRYABLE_API_CODES.has(error.code))
@@ -80,9 +87,11 @@ function isRetryableError(error) {
     }
     if (error && typeof error === "object" && "code" in error) {
         const code = String(error.code);
-        if (RETRYABLE_NETWORK_CODES.has(code))
+        if ((policy === "no-replay" ? NO_REPLAY_NETWORK_CODES : RETRYABLE_NETWORK_CODES).has(code))
             return true;
     }
+    if (policy === "no-replay")
+        return false;
     if (error instanceof Error && /timeout|ETIMEDOUT|ECONNRESET|socket hang up/i.test(error.message)) {
         return true;
     }
@@ -101,7 +110,7 @@ export async function withRetry(fn, options = {}) {
             return await fn();
         }
         catch (error) {
-            if (attempt >= retries || !isRetryableError(error))
+            if (attempt >= retries || !isRetryableError(error, options.policy ?? "default"))
                 throw error;
             // A server-sent Retry-After (429) wins over exponential backoff, but is capped
             // so it can't stall the CLI; otherwise fall back to jittered exponential backoff.

package/dist/src/version.js

@@ -1,2 +1,2 @@
 // Auto-generated — DO NOT EDIT
-export const CLI_VERSION = "0.24.0";
+export const CLI_VERSION = "0.26.0";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "gangtise-openapi-cli",
-  "version": "0.24.0",
+  "version": "0.26.0",
   "description": "CLI for Gangtise OpenAPI",
   "license": "MIT",
   "repository": {
@@ -23,6 +23,7 @@
     "README.md"
   ],
   "scripts": {
+    "prebuild": "node -e \"require('node:fs').rmSync('dist',{recursive:true,force:true})\"",
     "build": "tsc -p tsconfig.json",
     "dev": "tsx src/cli.ts",
     "test": "vitest run",
@@ -30,11 +31,11 @@
     "prepare": "node scripts/prepare.cjs && npm run build"
   },
   "engines": {
-    "node": ">=20"
+    "node": ">=20.18.1"
   },
   "dependencies": {
     "commander": "^14.0.0",
-    "undici": "^7.16.0"
+    "undici": "^7.28.0"
   },
   "devDependencies": {
     "@types/node": "^20.19.43",