Real-time PR crisis watchdog with Slack alerts and Notion log

Build me a real-time PR crisis watchdog as an agent workflow. The trigger is a cron schedule running every 30 minutes.

On each tick, the agent should query NewsWhip for articles mentioning my brand, my product names, and a configurable list of competitor names. Use NewsWhip's Quick Search Articles for fast last-24-hour scans, and Search Articles (Historical) when a richer Lucene-style query with engagement sorting is needed. Focus the search on the last hour, and only consider articles with high social velocity (NewsWhip's nw_score) or fb_overperforming signals above a configurable engagement threshold so low-noise mentions stay quiet.

For each candidate article, before doing anything else, the agent must dedupe against incidents already logged in Notion. Use Notion's Query a Database operation against my Crisis Log database, matched on the article URL, and skip any article that already has an entry. This is the rule that keeps the same story from re-alerting on subsequent ticks.

For each new high-signal article that survives dedupe, the agent should assess severity by combining (1) volume spike vs. baseline, (2) sentiment of the headline and snippet, and (3) publisher reach. Map that to a response level: Monitor, Prepare Response, or Escalate.

Then it should post a structured alert to a designated Slack channel using Slack's Send a Message operation. The Slack message must include: headline, publisher, link, key engagement metrics (nw_score, fb_overperforming, total shares), the assessed severity, and the suggested response level. Format it so the on-call comms person can scan it in two seconds.

At the same time, the agent should create a new page in my Notion Crisis Log database using Notion's Create a Page operation. The page should carry the headline as the title, plus properties for publisher, article URL, first-seen timestamp, engagement metrics, severity, and suggested response level. Leave a Timeline Notes section in the body so the comms team can append updates as the story develops. This is what keeps the durable incident record auditable after the Slack ping fades.

Configurable inputs the agent should expose: brand keywords, product keywords, competitor keywords, engagement threshold (nw_score floor and/or fb_overperforming floor), Slack channel ID, and Notion Crisis Log database ID. Default the schedule to every 30 minutes but make it easy to switch to every 15 or every 60.

Integrations to use: NewsWhip (Quick Search Articles, Search Articles (Historical)), Slack (Send a Message), Notion (Query a Database for dedupe, Create a Page for the incident record).