mirror of
https://git.openapi.site/https://github.com/desirecore/market.git
synced 2026-09-05 17:03:49 +08:00
Yige
aec2e7c28b
## 概述 / Overview 两件事：新增「企业微信助手」Agent（自带 15 个技能），并修正 `wecom-cli` 条目钉在 6 月快照的 ref 漂移。 Two changes: adds the **WeCom Assistant** agent (bundling 15 skills), and fixes the `wecom-cli` entry whose pinned ref was stuck on a June snapshot. ## 1. 新增企业微信助手 Agent 覆盖企业微信 **14 类服务、95 个方法**：消息、群聊历史、通讯录、日程、会议、待办、邮件、在线文档、在线表格、智能表格、智能文档、文档管理、微盘、媒体文件。 **采用内联形态 + 自带私有技能**：Agent 安装对 `agents/<id>/` 整目录递归复制且 `skills/` 不在排除集合里，因此装 Agent 即带全部技能，用户无需再单独获取技能合集。 ### 技能集（15 个，约 5000 行） - 基于上游 [wecom-cli](https://github.com/WecomTeam/wecom-cli) 官方 Skill（MIT，© WecomTeam）改写，每个技能末尾保留归属声明 - **新增 `wecom-chat`**：补齐上游零覆盖的群聊历史读取 - 补齐上游未覆盖的 `message.send`、`doc.create`，方法覆盖达 **95/95** - 修正上游三处文档漂移：邮件能力描述与实际相反、会议室参数名已过时、`title_highlight` 字段不存在 ### 相对上游的核心增量：风险治理 - 26 个对外可见或不可逆的方法逐个写明**执行前确认要求** - 4 个条件升级方法给出**参数级判据**，而非按方法名一刀切 - 文档权限扩散两项加重处理，涉及**企业外可见**时单独再确认一次 - 内部标识禁止外露，不因用户索要而放宽 - 拒绝导出可识别到具体自然人的隐私字段 ### 三条真机实测得出、上游未覆盖的硬约束 1. 机器人**只能写入/修改自己创建的数据**，真人创建的只能读 2. 每次响应携带的 `extra_identity_context` **禁止透露给用户** 3. 权限错误（`850002`/`851008`/`853006`）**不得重试**，须将 `help_message` **逐字原样**转给用户 ## 2. 修正 wecom-cli 条目 ref 漂移 `source.ref` 原钉在 2026-06-28 的 `72e14f7`，该快照只有 7 个子技能且用已废弃的旧命名（`msg`/`schedule`）。上游 v1.2.0 已扩展到 **14 个**技能。按旧 ref 安装的用户拿到的是三个月前的快照。 - `source.ref` → `78c514b2afee7c0d3d7be715628478421f37ee63` - `children` 由 `scripts/gen-collection-children.py` 重新生成，**7 → 14** - sidecar 同步 `provenance.content.ref`、`childCount` 与 `children` ## 验证 / Verification **静态** - 215 条示例命令追加 `--dry-run` 实跑，**215/215 退出码 0** - 未知方法 0、未知参数 0、`--json` 未知字段 0、枚举违规 0 - 15 个 `SKILL.md` 的 frontmatter 经客户端 `skillFrontmatterSchema` 校验全部通过 - `validate_catalog_metadata.py --require-complete` 与 `gen-collection-children.py`：**0 error** **真机（在真实企业微信账号上端到端）** - **待办域 6/6 方法全通**（含 2 个 write-high），`items` 必填的隐蔽坑实测证实 - **日程域 5 个方法全通**（含 3 个 write-high） - 消息发送、通讯录解析、微盘列表、邮件搜索、文档搜索、会议列表、智能表格创建均已实测通过 - 测试数据已全部清理，未污染真实账号 **尚未实测**：群聊历史（机器人未开通该品类）。相关文档已明确标注验证状态，未实测的能力不写「实际效果」段落。
67 lines
6.9 KiB
JSON
67 lines
6.9 KiB
JSON
{
|
||
"id": "wecom-assistant",
|
||
"name": "企业微信助手",
|
||
"avatar": {
|
||
"t": "微",
|
||
"bg": "linear-gradient(135deg, #34C759, #248A3D)"
|
||
},
|
||
"category": "communication",
|
||
"version": "1.0.0",
|
||
"updatedAt": "2026-09-03",
|
||
"maintainer": {
|
||
"name": "DesireCore Official",
|
||
"verified": true
|
||
},
|
||
"i18n": {
|
||
"default_locale": "en-US",
|
||
"source_locale": "zh-CN",
|
||
"locales": ["zh-CN", "en-US"],
|
||
"zh-CN": {
|
||
"name": "企业微信助手",
|
||
"shortDesc": "在对话里代你操作企业微信：消息、文档、表格、日程、会议、待办、邮件、微盘",
|
||
"fullDesc": "企业微信助手把企业微信的日常办公搬进 DesireCore 的对话框。你用日常语言说出意图，它通过企业微信官方命令行工具 wecom-cli 把事情办成，再用可读的语言汇报结果——不必打开企业微信客户端，不必记接口，不必自己敲命令。\n\n覆盖能力（14 类服务，95 个方法）：\n- 消息与会话：查最近会话、拉取群聊记录、发送文本/图片/文件/语音/视频\n- 在线文档：新建、导入、读取、追加与覆盖正文\n- 在线表格：新建、导入、读改数据、追加行、子表管理\n- 智能表格：子表/字段/记录/视图/图表的完整增删改查与行列样式\n- 智能文档：创建、页面读取与管理、Block 编辑、内容追加与覆盖\n- 文档管理：跨类型搜索、重命名、成员权限与加入规则\n- 日程：增删改查、参与人管理、多人闲忙查询、会议室查询与预订\n- 会议：预约、查询、搜索、取消、更新参会人、读取纪要与转写\n- 待办：创建、查询、更新、完成、删除、分派参与人\n- 邮件：发送、回复、转发、搜索与正文读取\n- 微盘：搜索、上传、下载、重命名、读元信息、新建文件夹\n- 通讯录：按姓名/拼音/别名解析成员\n- 媒体文件：本地文件与企微之间的上传下载\n\n风险治理：\n发消息、发邮件、改文档权限、覆盖或删除内容等 26 个对外可见或不可逆的操作，一律先向你复述影响并取得明确同意才执行。文档加入规则涉及企业外可见时会额外提示。所有内部标识（成员 ID、会话 ID、文档 ID 等）只在内部流转，回复中始终使用姓名、群名、文档标题这类可读信息。\n\n能力边界（真机实测得出）：\n机器人可以读取你的数据，但只能写入或修改机器人自己创建的内容——你自己建的文档、日程、待办，助手改不了，它会说明这条边界并给出替代方案。企业微信的机器人权限按品类逐项开通，未开通时助手会把官方开通指引原样转给你，不会反复重试。\n\n使用前提：\n需要 Node.js 18+ 与企业微信账号。首次使用时助手会引导你安装 wecom-cli 并用企业微信扫码完成授权，仅需一次。",
|
||
"tags": ["企业微信", "办公", "协作", "文档", "日程"],
|
||
"persona": {
|
||
"role": "企业微信办公助手",
|
||
"traits": ["稳妥确认", "说人话不露 ID", "覆盖全业务", "失败如实报告"]
|
||
}
|
||
},
|
||
"en-US": {
|
||
"name": "WeCom Assistant",
|
||
"shortDesc": "Operate WeCom from chat: messages, docs, sheets, calendar, meetings, todos, mail, drive",
|
||
"fullDesc": "WeCom Assistant brings everyday WeCom (Enterprise WeChat) work into the DesireCore chat box. You state your intent in plain language; it gets the job done through the official WeCom CLI and reports back in readable terms — no need to open the WeCom client, memorize APIs, or type commands yourself.\n\nCoverage (14 services, 95 methods):\n- Messaging: list recent sessions, pull group chat history, send text/image/file/voice/video\n- Docs: create, import, read, append and overwrite content\n- Sheets: create, import, read/modify data, append rows, manage subsheets\n- Smart sheets: full CRUD over subsheets, fields, records, views and charts, plus row/column styling\n- Smart pages: create, read and manage pages, edit blocks, append and overwrite content\n- Doc management: cross-type search, rename, member permissions and join rules\n- Calendar: full schedule CRUD, attendee management, multi-member free/busy, meeting room booking\n- Meetings: book, query, search, cancel, update attendees, read minutes and transcripts\n- Todos: create, query, update, finish, delete, assign participants\n- Mail: send, reply, forward, search and read message bodies\n- Drive: search, upload, download, rename, read metadata, create folders\n- Contacts: resolve members by name, pinyin or alias\n- Media: move files between your machine and WeCom\n\nRisk governance:\n26 operations that are externally visible or irreversible — sending messages or mail, changing document permissions, overwriting or deleting content — always restate their impact and require your explicit consent first. Join-rule changes that would expose a document outside the company get an extra warning. Internal identifiers never appear in replies; you always see names, group titles and document titles.\n\nCapability boundary (verified on a live account):\nThe bot can read your data but may only write or modify content the bot itself created — documents, schedules and todos you created yourself cannot be modified; the assistant explains this boundary and offers an alternative. WeCom bot permissions are granted per category; when a category is not enabled the assistant relays the official activation guidance verbatim instead of retrying.\n\nRequirements:\nNode.js 18+ and a WeCom account. On first use the assistant walks you through installing wecom-cli and authorizing once by scanning a QR code in WeCom.",
|
||
"tags": ["wecom", "office", "collaboration", "documents", "calendar"],
|
||
"persona": {
|
||
"role": "WeCom office assistant",
|
||
"traits": ["confirms before acting", "plain language, no raw IDs", "full business coverage", "reports failures honestly"]
|
||
},
|
||
"translated_by": "human"
|
||
}
|
||
},
|
||
"persona": {
|
||
"tools": []
|
||
},
|
||
"changelog": [
|
||
{
|
||
"version": "1.0.0",
|
||
"date": "2026-09-03",
|
||
"changes": {
|
||
"zh-CN": [
|
||
"首次发布：覆盖企业微信 14 类服务、95 个方法",
|
||
"内置 15 个技能，随 Agent 一并安装，无需单独获取",
|
||
"对 26 个高风险操作实施执行前确认",
|
||
"全流程禁止外露内部标识，回复统一使用可读名称",
|
||
"待办与日程两个业务域已在真实企业微信账号上端到端验证"
|
||
],
|
||
"en-US": [
|
||
"Initial release: covers 14 WeCom services and 95 methods",
|
||
"Ships 15 bundled skills installed together with the agent",
|
||
"Pre-execution confirmation for 26 high-risk operations",
|
||
"Internal identifiers never surface in replies; readable names throughout",
|
||
"Todo and calendar domains verified end-to-end on a live WeCom account"
|
||
]
|
||
}
|
||
}
|
||
]
|
||
}
|