PL-10 Sample v1.0 · updated 15 Aug 2026
Extract fields from text into a table, marking what is missing
Pulls the fields you name out of emails, forms or notes into one table, and writes missing instead of guessing a value.
- Task
- Analyse
- Level
- Practitioner
- Placeholders
- 3
- Works with
- Chat, API, Local
Extract these fields from each record in the text below: {{fields}}.
Rules:
- One row per record. Records are separated by: {{separator}}.
- Copy values exactly as written; do not reformat or correct them.
- If a field is not stated, write MISSING. Never guess.
- If a value is ambiguous, write AMBIGUOUS and add a note after the table.
- Output a table with one column per field, in the order given, then a column called Source quoting the words you used.
Text:
"""
{{text}}
"""
Words in double braces are placeholders. Replace them before you send the prompt, here or in your tool.
Fill in the prompt
Filling in on the page needs JavaScript. Copy the prompt by selecting it, then replace each placeholder in your tool.
| Placeholder | What to put there | Example |
|---|---|---|
{{fields}} | Fields to extract. Comma-separated, in the column order you want. | name of the organisation, event date, number of attendees |
{{separator}} | How records are separated. Describe it so the model can split the text. | a line of three dashes |
{{text}} | The text. All records together. | (paste the records here) |
Is it the right prompt?
Use it when
- The same few facts are buried in many similar messages.
- You will paste the result into a spreadsheet.
Skip it when
- Hundreds of records: write a script or use a tool built for extraction.
- The text holds personal data and your tool is not approved for it.
How to adapt it
- Values need a standard format
- Add a separate step after extraction, so the original values are kept for checking.
- Records are long
- Send them in batches of ten and merge the tables.
Check the answer
Before you rely on the result, confirm that:
- The number of rows matches the number of records.
- MISSING appears where the text really does not say, and nowhere else.
- A random sample of values matches the Source quotes.
Watch out for
- Accuracy Check a sample of rows against the source before using the table for anything that matters.
- Privacy Extraction concentrates personal data in one place. Store the result as carefully as the source.
Revision history
| Version | Date | Change |
|---|---|---|
| v1.0 | First sample version. |