Lets the Agent query Chen Yiwei's (Moderator Chen) practical Q&A data on the Accounting Vision Forum in the most natural Chinese, without opening a browser. SKILL.md standard format, usable across any Agent platform such as Claude Code / Codex CLI / Cursor / Gemini CLI / OpenCode.
Online:https://bbs.auditdog.cn(公开匿名可访,无需 token)
When to use
Use this Skill when the user asks the following types of audit/accounting practical questions:
- "What does Moderator Chen say about XXX"
- "Practical treatment of accounting standard XXX"
- "How to do XXX in audit practice"
- "Discussion about XXX on the Vision Forum"
- "What Q&As does Moderator Chen have recently"
- "Moderator Chen's latest answers"
- "CPA audit XXX"
Endpoint overview
| Endpoint | Purpose | Main parameters |
|---|---|---|
/api/public/recent |
Recent Q&A (by time) | days (1-3) / page (pagination, 20 per page) |
/api/public/search |
Keyword + semantic search | q (keyword, at least 2 characters) / page (pagination, 10 per page) / scope (optional: all/title/question/comment) |
/api/public/detail |
Full Q&A content of a post | link (post link) |
Conventions:
- Base URL:
https://bbs.auditdog.cn - Authentication: none (anonymous)
- Rate limit: 30 req/min/IP
daysupper limit 3 days, server-side hard limit protectionscopeoptional, when not passed, search byallfull search (title + question + replies)- Automatically triggered when keyword hits are insufficientsemantic search(posts that do not overlap in wording but are similar in meaning), no need for the frontend to be aware
Workflow
Fetch recent Q&A (user asks "what Q&A does Moderator Chen have recently")
# 拉最近 1 天的问答(第 1 页)
curl -s "https://bbs.auditdog.cn/api/public/recent"
# 拉最近 3 天的问答
curl -s "https://bbs.auditdog.cn/api/public/recent?days=3"
# 翻页(第 2 页)
curl -s "https://bbs.auditdog.cn/api/public/recent?days=3&page=2"
Keyword search (user asks "what does Moderator Chen say about revenue recognition")
# 搜索关键词(第 1 页)
curl -s "https://bbs.auditdog.cn/api/public/search?q=%E6%94%B6%E5%85%A5%E7%A1%AE%E8%AE%A4"
# 搜索多个关键词(空格分隔,AND 逻辑)
curl -s "https://bbs.auditdog.cn/api/public/search?q=%E5%90%88%E5%B9%B6%E6%8A%A5%E8%A1%A8+%E5%86%85%E9%83%A8%E4%BA%A4%E6%98%93"
# 翻页(第 2 页)
curl -s "https://bbs.auditdog.cn/api/public/search?q=%E6%94%B6%E5%85%A5%E7%A1%AE%E8%AE%A4&page=2"
View full content of a post (user wants to see the detailed Q&A of a post)
# 用搜索或最近接口拿到的 link 字段,URL encode 后传入
curl -s "https://bbs.auditdog.cn/api/public/detail?link=$(python3 -c 'import urllib.parse; print(urllib.parse.quote("https://bbs.esnai.cn/thread-5497367-1-1.html"))')"
Typical usage flow
User asks: "What does Moderator Chen say about revenue recognition?"
- First search by keyword:
/api/public/search?q=收入确认 - After finding the relevant post, use the
linkfield to fetch the full content:/api/public/detail?link=... - Organize the Q&A content and present it to the user
User asks: "What new Q&A does Moderator Chen have recently?"/"Moderator Chen Q&A daily report"
- Fetch the last 1 day:
/api/public/recent(already includes full Q&A content) - Directly organize into daily report format and present to the user
User asks: "What did Moderator Chen answer in the last 3 days?"
- Fetch the last 3 days:
/api/public/recent?days=3 - Organize and present
User asks: "What exactly did the first post say?"
- Use the
linkfrom the list to call/api/public/detail - Display the full question and reply content
Returned data format
/api/public/recent returns
{
"days": 1,
"page": 1,
"pageSize": 20,
"total": 35,
"count": 20,
"hasMore": true,
"results": [
{
"title": "关于XXX的会计处理",
"link": "https://bbs.esnai.com/thread-XXX-1.html",
"reply_count": 5,
"latest_time": "2026-05-12T10:30:00.000Z",
"posts": [
{
"pid": "12345",
"question_author": "审计小白",
"question_time": "2026-05-12 09:30:00",
"question_text": "完整的问题内容(纯文本)...",
"author": "陈版主",
"comment_time": "2026-05-12 10:15:00",
"comment_text": "完整的回复内容(纯文本)..."
}
]
}
]
}
/api/public/search returns
{
"keyword": "收入确认",
"page": 1,
"pageSize": 10,
"count": 10,
"hasMore": true,
"scope": "all",
"semantic_triggered": false,
"semantic_affected": false,
"results": [
{
"title": "关于收入确认时点的问题",
"link": "https://bbs.esnai.com/thread-XXX-1.html",
"reply_count": 3,
"latest_time": "2026-05-10T08:00:00.000Z",
"preview": "摘要文本...",
"match_type": "keyword"
}
]
}
Notes:
scope: Limits the search scope (all/title/question/comment); if not provided, defaults to allsemantic_triggered: Whether semantic search was triggered this time (only triggered when keyword hits are insufficient)- Each
match_type:keyword(keyword hit) orsemantic(semantically related recommendation; the wording may not contain the query term) - If the user asks using colloquial/synonymous wording,
match_typemay besemantic(a semantically recalled post), and its body may not contain the query term verbatim—in this case, you should tell the user that this is a "semantically related recommendation" rather than an exact hit
/api/public/detail returns
{
"title": "关于收入确认时点的问题",
"link": "https://bbs.esnai.com/thread-XXX-1.html",
"posts": [
{
"pid": "12345",
"question_author": "审计小白",
"question_time": "2026-05-10 09:30:00",
"question_text": "完整的问题内容(纯文本)...",
"author": "陈版主",
"comment_time": "2026-05-10 10:15:00",
"comment_text": "完整的回复内容(纯文本)..."
}
]
}
posts The array is sorted by time; a post may contain multiple rounds of Q&A.question_text is the question content,comment_text is Moderator Chen's reply.
Output format for the user
Recent Q&A / search results list
**陈版主实务问答 — 最近 1 天**(共 8 条)
1. **关于XXX的会计处理** — 3 条回复 · 2小时前
🔗 https://bbs.esnai.com/thread-XXX-1.html
2. **关于YYY的审计程序** — 5 条回复 · 昨天
🔗 https://bbs.esnai.com/thread-YYY-1.html
hasMore=true , prompt the user "there is more, you can say 'next page' to continue viewing".
Moderator Chen Q&A Daily Report (when the user says "daily report"/"what Q&A is there today")
When the user asks for the "daily report" or "today's Q&A", use /api/public/recent?days=1 to get the complete content, then organize it into daily report format:
**陈版主答疑日报 · 2026-05-12**(共 8 个问题)
---
**1. 关于XXX的会计处理**
💬 提问:完整问题内容...
✅ 陈版主回复:完整回复内容...
---
**2. 关于YYY的审计程序**
💬 提问:完整问题内容...
✅ 陈版主回复:完整回复内容...
---
*数据来源:bbs.auditdog.cn · 会计视野论坛*
Post details (when using the detail endpoint)
**关于收入确认时点的问题**
💬 **审计小白** · 2026-05-10 09:30
完整的问题内容...
---
✅ **陈版主** · 2026-05-10 10:15
完整的回复内容...
---
💬 **审计小白** · 2026-05-10 11:00
追问内容...
---
✅ **陈版主** · 2026-05-10 11:30
再次回复内容...
🔗 原帖:https://bbs.esnai.com/thread-XXX-1.html
Output requirements
- Time must be converted into human-friendly language: "2 hours ago", "yesterday", "3 days ago"; do not directly display ISO timestamps
- When the user wants to see the complete content of a post, proactively call
/api/public/detailto fetch it; do not stop at preview only linkis the original post URL on the Accounting Vision Forum; users can visit it directlyreply_countis the number of replies to that post- Do not expose technical details such as endpoint paths, parameter names, or rate-limit values in user output
Common error handling
{"error":"请求过于频繁,请稍后再试"}(HTTP 429): rate limit triggered; wait 1 minute and retry{"error":"请输入搜索关键词"}(HTTP 400): missingqparameter{"error":"关键词至少2个字符"}(HTTP 400): search term is too short{"error":"缺少 link 参数"}(HTTP 400): the detail interface requires a link parameter{"error":"未找到该帖子"}(HTTP 404): link is invalid or the post does not existresultsis an empty array: no matching Q&A; suggest the user change keywords
Do not do
- Do not guess or fabricate Q&A content — always rely on what the API returns
- Do not poll at high frequency — data updates daily, and the same question does not need repeated calls
- Do not stop after only showing preview — when the user asks for specific content, call the detail interface
- Do not expose technical details such as endpoint paths, parameter names, or rate-limit values in user output
- Do not aggressively pull pages concurrently — serial + natural intervals
Author: nigo WeChat public account: 逆行的狗