First CH Apps / LINE Attendance Tally
When attendance replies came in as free text instead of a LINE poll, paste the exported chat history to count going / not going / not sure and head counts (○2), and match it against your member list to see who has not replied.
When you take attendance in LINE, everyone replies in their own way and counting takes time. This page counts the text replies just by pasting the chat back. Fix the reply format on one line in your next call and there will be fewer rows to check.
No. Reading and counting happen on the device you opened this page on. Something is sent only when you turn on the AI option, review what will be sent and allow it.
If you use a poll or event from the start, it counts for you. This page is for when replies came in as text instead: paste the chat history back to count them, see who has not replied against your list, add up head counts like “○2” and export CSV.
Only the text from LINE’s Export chat history (date lines and “time, name, message” lines) is supported. Messages copied by long-pressing in the chat lack the dates and separators. After you paste, what is missing (date lines, time / name separators) is shown below the box.
One event per tally. If there are several 【出欠確認】 messages, the one matching the event name in step 1 (or the first one) is used.
Writing the call, reading the chat history, classifying, counting and creating the summary, reminder and CSV all happen on this device. The chat history, member list and event details are never sent to the server.
The only thing sent is, when you turn on Use the AI option and allow it after reviewing, the text of the To check rows you picked and the message before each (with speakers’ and listed names, phone numbers, e-mail addresses, @mentions and names before さん etc. masked; the server masks phone numbers, e-mail addresses, @mentions and names before さん etc. again. Names not on your list may remain, so you can review exactly what will be sent first). Our server (Cloudflare Pages Functions) passes it to Cloudflare Workers AI and returns the result.
First CH does not store the text or the result (nothing is written to a database or log, and it is gone when the request ends). Cloudflare handles it under its official policy.
The database only records the date and the number of AI uses that day (the total across all users and the count for this tool). To prevent rapid repeats, requests per IP are counted in server memory only, never stored, and cleared within 10 minutes.
POST /en/henji-kuru/api/read with JSON ({"lines": [{"text": "私も", "context": "○2"}], "count": true}) returns each row's category (yes / no / maybe / unclear), head count and reason as JSON (lines: 1–10; text and context up to 300 characters each; count: true if the call asked for numbers. No auth, nothing stored; the daily and per-IP limits are shared with the page). Tallying a whole chat runs on your device, so there is no API for it.