Monday morning foot-traffic digest for your shopping center portfolio

Build me an agent workflow that runs on a cron every Monday at 7am local time and posts a weekly foot-traffic momentum digest to a Slack channel for my shopping center portfolio.

Inputs the agent should read at the top of every run:

1) A Google Sheet that lists my portfolio locations. Use Google Sheets Get Values to read it. The sheet has one row per location, with one column for the Sites USA shopping center ID and one column for the merchant Placekey for the specific store inside that center. Include extra label columns like location name or region if present, and use them as display names in the digest. The spreadsheet ID, sheet/tab name, and column mapping should be configurable inputs.

2) A configurable drop threshold for the alert section, defaulting to 15 percent.

3) The target Slack channel ID for the digest.

For each row in the sheet, call Sites USA Get Store Visits by Shopping Center using the shopping center ID, and Sites USA Get Store Visits by Placekey using the Placekey, to retrieve the trailing twelve months of monthly visit data for both the center and the specific store. Pace requests to respect the Sites USA 1 call per second limit.

For each location, compute month-over-month change (most recent complete month vs the prior month) and year-over-year change (most recent complete month vs the same month last year), as both absolute visit counts and percent change. Do this at both the shopping center level and the store/Placekey level so the digest can show both.

Then build the digest with these sections, in this order:

- A one-line headline with the reporting period (the most recent complete month) and a quick portfolio-wide read (e.g. how many locations are up year-over-year vs down).

- Top 3 risers, ranked by year-over-year percent change. For each, show the location label, the year-over-year delta, the month-over-month delta, and a one-sentence agent interpretation of what is likely driving the move.

- Top 3 fallers, same shape as risers.

- A "watchlist" section listing every location whose year-over-year drop exceeds the configurable threshold (default 15 percent), even if they are not in the top 3 fallers. Same per-line format with a one-sentence interpretation.

- A short footer titled "No data this cycle" listing any locations where either Sites USA call returned no visit data, so nothing gets silently dropped.

Post the assembled digest to the configured Slack channel using Slack Send a Message. Format with Slack mrkdwn: use *bold* (single asterisks) for section headers and location names, and keep each location bullet to one or two compact lines. Round percentages to whole numbers and prefix with + or - so direction is obvious at a glance.

Operational notes: handle Sites USA 401 by re-authenticating (the credential is an API key that exchanges for a JWT good for 12 hours). Treat empty arrays from either Sites USA call as "no data" and route the location to the footer rather than failing the run. If the Slack post fails, retry once before erroring.