Skip to content
FlintFlintby Torch
← LevelsBuilder · B11

Sheet-connected pipelines

Build-along · ~12 min at your own keyboard

The pipeline pattern: a workflow gathers or computes, and its last step writes rows into a sheet (via API or file). The sheet stays the interface your business already knows; the filling becomes invisible. The hard part is identity: what makes two rows 'the same record'? An email isn't — one client can have five jobs. Decide what identifies a record (a job ID, or job ID + date), what happens when a known record comes back changed, and test the awkward cases, not just a rerun.

The long version: Pipelines into a sheet ↗

Do this

0/4

Checkpoints

  • ▸You should now see new rows added to your sheet or CSV with a run-date.
  • ▸You should now see your identity and change rules written down.
  • ▸You should now see results for all four test cases.
  • ▸You should now see a warning when you feed it too few rows.
The three ways this goes wrong
  • The second run adds the same rows again, or a client's second job disappears. — Your identity rule is wrong. Use a field that's unique per record, like a job ID, not one shared across records, like an email.
  • The run-date column is empty. — Ask Claude to write today's date on every appended row.
  • Rows are missing and nothing says so. — Add the threshold rule: fewer rows than expected means warn and do not write.

Share what happened

Paste your pipeline description, your identity and change rules, the four test results, and the threshold rule.

Strip API keys, passwords, and tokens before you paste. Swap them for [KEY].

🔒 Don't paste confidential, personal-health, or payment info.