Social Listening Monitor
Search public conversations with SocialListeningAPI, then turn the returned posts into a
sourced review list. Treat search results as research leads. A person must review them before
acting.
Run a search
-
Ask which query to search and which of linkedin, x, and reddit to include.
-
Run this fixed command through the exec tool:
node "{baseDir}/scripts/search-social.js"
-
Pass user input through the exec tool's environment map, never through the command string:
SOCIALLISTENING_QUERY=
SOCIALLISTENING_PLATFORMS=linkedin,x,reddit
-
Let the command inherit SOCIALLISTENING_API_KEY from the OpenClaw process.
-
Review errors before summarizing. Keep usable results when one source fails.
-
Cite every finding with its item url. Never invent missing context or results.
The script returns normalized JSON. items are sorted newest first. creditsUsed reports the
credits returned by successful API calls. sourceUrls contains the public post URLs.
Choose the workflow
- For a brand review, search the exact brand name and likely misspellings separately.
- For competitor research, search focused phrases about pricing, launches, switching, and
problems. Avoid one broad query that mixes different questions.
- For a weekly brief, search the brand and competitor, remove URLs already present in the
user's own history, group the remaining posts by theme, and cite each point.
Read references/monitoring-workflows.md for query patterns and brief structure.
Limits
SocialListeningAPI performs public search. It does not schedule runs, store history, remove
duplicates, classify posts, send Slack messages, or create alerts. Use OpenClaw or another
external tool for those steps. Search results can include irrelevant posts, so require human
review.