Build a data export with AI: scope, permissions and reproducible files

A worked example: where the task becomes difficult

A fictional account dashboard exports order history for bookkeeping. The visible table shows one filtered page, but the user expects all matching orders. Another account’s records must never enter the file. Clarify whether export means current page, current filter or complete account history, then use controlled records containing commas, quotes and multiline notes to inspect output.

Decisions to make before implementation

Choose permitted fields, ordering, filter interpretation and format. Keep private operational columns out unless explicitly required. Decide how large exports run and how the user retrieves an accepted job. Spreadsheet-bound text may need handling so untrusted cell content is not treated as executable formulas. Preserve source timestamps and units so exported amounts remain interpretable outside the app.

A practical sequence for the work

Use the sequence below as a task boundary, not as a claim that the example has been executed. Work with approved inputs and the project’s actual architecture. If a required integration or permission is unavailable, keep that stage visibly incomplete rather than generating a plausible substitute result.

• Write the export scope and field dictionary from the customer task.
• Implement authorized data selection independently of browser table visibility.
• Serialize representative values with an appropriate format and safe text handling.
• Inspect the downloaded file and large-export recovery behavior.

A detailed brief you can adapt for your agent

Replace the illustrative context with your approved facts and controlled inputs. Keep the stated boundaries when adapting the brief. The expected deliverable matters more than a particular tool name: ask for an explanation grounded in the inspected material and evidence for the requested outcome.

Build an order export for the approved account and filter scope. Select data through an authorized receiving operation, not a complete cross-account browser download. Define fields, units and ordering. Handle quoted text and spreadsheet-sensitive values appropriately. Verify the actual file with controlled records and protect large-export retrieval.

Failure modes that an attractive preview can hide

Exporting only currently loaded rows can silently omit most history. Fetching every account’s rows and filtering in the browser creates unnecessary exposure. Do not place private exports at permanently public URLs. A generated filename and success toast do not prove that the file contains the requested records or remains accessible only to its owner.

Technical references: MDN: form validation

Acceptance checks and the evidence to retain

Keep the scope definition, field dictionary, authorization evidence and parsed-file checks. Report export-job acceptance separately from completed file availability.

Specific answers

Common questions