鱼泡求职 MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| YUPAO_DATA_DIR | No | Directory where the local SQLite workspace is stored. The candidate pool is saved at YUPAO_DATA_DIR/workspace.sqlite3. | .local |
| YUPAO_CHAT_BROWSER | No | Browser used by the optional chat SDK integration. Defaults to installed Chrome. Set to msedge to use Edge, or chromium after installing its Playwright browser runtime. | chrome |
| YUPAO_SESSION_FILE | No | Path to a locally saved login JSON file containing token and userrole. Job seeker userrole is 2; recruiter is 1, and recruiter identity is rejected. Do not commit this file or copy it into chat. | |
| YUPAO_CHAT_SDK_FILE | No | Path to the downloaded fixed-version Tencent Cloud official browser SDK entry file, e.g. .cache/chat-3.6.7/index.js. Required only for reading conversations and chat records via the optional chat SDK integration. | |
| YUPAO_ENABLE_WRITES | No | Real writes are disabled by default. Set to 1 to enable the restricted single-action executor. This only enables the executor and does not approve any target. | unset |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| yupao_capabilitiesA | 查看已实现、已验证及尚未启用的能力，不请求鱼泡。 |
| yupao_search_jobsA | 按关键词与城市实时搜索鱼泡岗位，分页返回结构化结果。不会投递或打招呼。 |
| yupao_get_job_detailA | 根据搜索结果的 job_biz_id 读取完整 JD，供调用方 AI 进行岗位分析。 |
| yupao_check_sessionA | 检查明确配置的本机鱼泡登录文件，不读取浏览器凭证，不返回 token。 |
| yupao_list_resume_attachmentsA | 列出本人已上传的附件简历元数据。for_chat=true 用于聊天附件；不下载或发送简历。 |
| yupao_check_messaging_sessionA | 核对鱼泡能否提供 IM 登录资料，不返回凭证；未登录 SDK，不读取聊天或发送消息。 |
| yupao_list_conversationsA | 通过官方 SDK 同步会话并读取 HR 联系人资料；默认排除系统助手。需要 chat 依赖，不标记已读。 |
| yupao_get_messagesB | 分页读取已同步会话的聊天记录；不发送消息或已读回执，不返回附件下载地址。 |
| yupao_prepare_applicationB | 预览单条指定附件投递，显示发布者、简历及权限；准备阶段不发起沟通或投递。 |
| yupao_prepare_messageB | 预览单条消息；已有会话须核验关联，start_conversation=true 可预览免费首次开聊。准备阶段不发送。 |
| yupao_save_presetC | 保存搜索条件；薪资为元/月，学历与经验表示本人条件。平台未核实的过滤在本机执行。 |
| yupao_list_presetsA | 读取本机保存的搜索预设，不请求鱼泡。 |
| yupao_remove_presetC | 移除指定本机搜索预设，不操作平台。 |
| yupao_search_presetA | 按预设搜索并读取完整 JD 后本机筛选；每页最多 10 个详情，未知条件单列待核实。 |
| yupao_save_jobB | 读取完整 JD 保存到本机候选池，按 job_biz_id 去重，保留备注和标签。 |
| yupao_list_saved_jobsC | 按标签、状态读取本机候选池，不请求平台。 |
| yupao_annotate_jobC | 修改本机岗位标签、备注及 candidate/priority/excluded/archived 状态。 |
| yupao_remove_saved_jobA | 移除指定本机候选岗位，不撤销平台投递或沟通。 |
| yupao_prepare_matchB | 保存完整 JD 与本人提供的简历材料，生成供当前宿主 AI 分析的材料与引文要求；不另调模型。 |
| yupao_record_matchB | 保存宿主 AI 的逐项匹配分析；校验 JD/简历引文，区分未知与不满足并汇总权重。 |
| yupao_list_matchesB | 读取和排序本机匹配分析，标明保存的 JD 是否已更新；分数不是录用概率。 |
| yupao_get_actionA | 读取具体操作的完整预览、确认摘要、执行状态和回执，不执行任何发送。 |
| yupao_execute_actionA | 执行已预览的单条操作；必须由真人确认。没有自动批准参数；默认关闭，结果不明不重发。 |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 23 tools
Tools are mostly grouped into clear workflows: job search/detail, match preparation/recording, presets, saved jobs, sessions, messaging, and action preview/execution. A few boundaries require careful reading, especially check_session vs check_messaging_session, and prepare_message/prepare_application vs execute_action, but descriptions generally clarify them.
All tools share the yupao_ prefix and snake_case style, with near-uniform verb_noun naming such as search_jobs, get_job_detail, save_preset, and remove_saved_job. The noun-only capabilities tool is a minor exception but remains unambiguous.
The 23-tool surface covers many subdomains, and most tools are distinct, but it is heavy for an MCP server. Several CRUD/lifecycle groups, especially presets and saved jobs, could potentially be consolidated without losing much capability.
Coverage is broad: search, JD retrieval, matching, presets, saved jobs, sessions, conversations, message/application previews, and gated execution are all represented. Minor gaps remain, such as no explicit single-match retrieval/update path and limited resume attachment operations beyond listing metadata.