Lead List Export

Use this whenever leads have been fetched. A list of people is a spreadsheet, not a paragraph.

The rule

After runDataSource returns leads, call createDocument once with kind: "sheet" and pass the rows as CSV in content. Do not describe the list and wait to be asked for a file — the spreadsheet is the deliverable.

Column order

Use exactly this header row, every time, so exports stay diffable and importable:

Full Name,Job Title,Company,Company Size,Industry,Email,Personal Email,Mobile,LinkedIn,City,Country,Website

Map straight from the row fields: fullName, jobTitle, companyName, companySize, industry, email, personalEmail, mobile, linkedin, city, country, companyWebsite.

Copy verbatim — this is the part that matters

Every value is contact data somebody will actually send to. Transcribe it exactly as it came back.

• Never invent, complete, or "fix" an email address. Not a domain, not a spelling.
• Never guess a phone number or a LinkedIn URL.
• A null field becomes an empty cell. Not "N/A", not a placeholder, not a plausible-looking value.
• Do not reorder, dedupe, or drop rows unless asked. The sheet should hold what the search returned.

A single fabricated email in an exported list is worse than a missing column — it bounces, it damages the sender's domain, and it is invisible until it does.

CSV mechanics

• Quote any value containing a comma, quote, or newline; double an inner quote (" becomes "").
• Job titles and company names contain commas constantly. Quote them by default.
• One row per lead, no blank line between rows, no trailing commentary inside the file.

Title the sheet usefully

Name it after the search, not the format: Marketing Directors — France (25 leads), never "Leads" or "Spreadsheet".

After the sheet

The document ends your turn, so say everything you need to before calling createDocument: the count, the standouts, and any filterNote. Follow-ups like "now narrow this to Paris" start a fresh search.