Daily JSearch job tracker with Google Sheets log and Slack recap

Build a code workflow that runs every weekday at 7am America/New_York and keeps a Google Sheet up to date with fresh job postings from JSearch, then posts a Slack recap. Pure deterministic pipeline: read, fetch, diff, append, notify.

Trigger: cron, Monday through Friday at 07:00 ET.

Inputs the workflow should expose as configuration: the Google Sheets spreadsheet id, the name of the Queries tab (default "Queries"), the name of the Tracked tab (default "Tracked"), and the Slack channel id for the recap.

Step 1 — Read saved searches from Google Sheets using Get Values on the Queries tab. Each row holds a query string in column A (for example "senior backend engineer remote") plus optional filter columns: employment_types (full-time, part-time, contract, intern, comma-delimited) and date_posted (today, 3days, week, month). Skip the header row and skip any row with an empty query.

Step 2 — Read the existing tracked job ids from Google Sheets using Get Values on the Tracked tab, column A only (the job_id column). Build an in-memory set of ids already seen so we can dedupe.

Step 3 — For each row from step 1, call JSearch Search Jobs with that row's query and filters. Use page=1 and num_pages=1 unless overridden. Note that JSearch expects employment_types and date_posted as query string params, and the location should be embedded inline in the query string (the API has no separate location parameter). Collect each returned posting's job_id, job_title, employer_name, job_city/job_state/job_country, job_posted_at_datetime_utc, job_min_salary/job_max_salary/job_salary_currency/job_salary_period, and job_apply_link.

Step 4 — Dedupe. Drop any posting whose job_id is already in the set from step 2. Also dedupe across queries within this run so a posting that matches two saved searches is only logged once (keep it under the first query that matched, for the recap).

Step 5 — Append the new rows to the Tracked tab using Append Values with valueInputOption USER_ENTERED. Columns, in order: job_id, title, company, location (formatted "City, State, Country" with empty parts omitted), posted_at (ISO date), salary_range (formatted like "$120k–$150k / year" when min and max are present, otherwise the single value or empty), apply_url.

Step 6 — Post one Slack message via Slack Bot Send a Message to the configured channel. Format with Slack mrkdwn. Headline line: "Fresh roles for {weekday date}: {total new} new". Then one section per query that produced results, in the form "*{query}* — {count} new", followed by up to 5 bullet lines linking the top postings as "<{apply_url}|{title}> at {company}". Omit queries that returned zero new postings. If the entire run produced zero new postings across every query, still post a short "No new roles today" message so the channel knows the watcher is alive.

Error handling: if JSearch returns a 429 or 5xx for a particular query, log the failure and continue with the next query rather than aborting the whole run; surface failed queries at the bottom of the Slack recap as "Could not fetch: {query}".

Integrations to use: jsearch (Search Jobs), google-sheets (Get Values, Append Values), slackbot (Send a Message).