Keyword Monitor

Each keyword holds one search query and runs it against all of Reddit — not just the communities you watch — plus everything already in Prowlo's crawl corpus. Open Keyword Monitor in the dashboard sidebar and click New keyword to add one.

• First results in minutes — the first crawl is queued when you create it
• Checked about daily — each keyword is re-searched roughly once a day
• Up to 25 new matches per day per keyword are kept
• Per-keyword tags — attach tags (e.g. competitor) and every find is labeled with them automatically
• Matches are embedded into your Dataset — searchable by meaning, taggable, and alertable like any other record

A keyword's query supports three matching styles:

• invoice — broad: Reddit's own matching for the term
• "invoice generator" — exact phrase: keeps only posts containing exactly this
• invoic* — any ending: also finds "invoices", "invoicing"…

Each keyword can carry its own intake filters and tagging rules. Intake filters are a hard gate — items must pass every filter (minimum length, minimum score, language, post age) to be ingested at all, so noise is dropped before it reaches your Dataset. Tagging rules label matched records (with ALL/ANY conditions and a Preview matches button) so you can filter your Dataset to them — items that don't match a tag still ingest.

Keyword tagging rules ride along on community Watchers. As a Watcher ingests its channel, every post is checked against your terms in real time; matches are kept and tagged, and adding a new rule backfills matches from the last 7 days of ingested posts. Matching is stemming-aware — "deploy" also catches "deploying", "deployed", and "deployment". Tagging rules are uncapped on every plan.

• Keyword Monitor — when you don't know where the conversations happen: brand names, competitor names, niche category phrases. Broad coverage, daily cadence.
• Tagging rule — when you already watch the right communities and want matches labeled within them: pain-point phrases, feature terms. Deep coverage, real-time during ingestion.
• Both together — a keyword finds which communities talk about you; community Watchers then go deep on the best ones with tagging rules.

Under the hood each keyword is a Watcher holding one reddit_search source, so your agent creates one with watcher_create, using the query as the channel:

watcher_create {
  "sources": [
    { "subreddit": "CRM alternative", "platform": "reddit_search" }
  ]
}

Tagging rules use the keyword_* tools (keyword_list, keyword_create, keyword_update, keyword_delete). Or just ask: "Track 'CRM alternative' across all of Reddit" or "Add a tag for 'onboarding frustration'".

• Search — matches live in your Dataset; query them by keyword or by meaning (see POST /v1/search)
• Alerts — fire a Slack message or webhook when a new match lands (see Alerts & Trends)
• Trends — chart how often a term appears over time