
Build rich applications with real-time social data. Powered by @LunarCrush.
California, United States
Joined July 2019
- Tweets13.1K
- Following4
- Followers15.7K
- Likes7.4K
most monitoring RFPs list networks and seats. neither predicts the question you get asked in March.
/categories/list/v1 came back with 41 categories this morning, Formula 1 next to technology brands, six networks under each.
you will use four. a trial finds out which four.
legal has a definition of your brand. the topics list has a different one.
/searches/create takes yours and returns a slug that runs anywhere a topic does: time-series, posts, creators.
the dashboard still wants volume. one of them is writing your search.
where custom searches sit in the endpoint reference: lunarcrush.com/en/developers…
lunarcrush /public/topic/samsung/v1
the passthrough sends any catalog path through, auth already handled. no client library.
this morning: Samsung engagements up 10% on the week, down 13% on the month.
so is Samsung up? depends which row your exec read.
install is one line, and it ships with every paid plan: lunarcrush.com/en/developers…
Resident Evil sentiment reads 16% this morning. nothing happened.
it has sat at 17% for six months. a 50% alarm here never stops ringing.
sentiment weights posts by interactions. my read: horror vocabulary scores negative before anyone has an opinion.
best day in a year: 42%.
we counted what a naive 2x rule would have paged you for this week.
Anthropic, hourly. baseline near 3.4M engagements an hour, line at twice that.
45 of 168 rows cleared it. the chart has two events.
how long is your cooldown?
your clipping service sees 72 posts.
Sony, last 24 hours as of this morning: 30,296 mentions across six networks. 72 of them are news.
that is 0.11% of the engagements, and the only slice most comms decks contain.
/topic/sony/news/v1 returns it on its own.
what the news stream gives you for any topic, pulled on its own: lunarcrush.com/en/developers…
is your workload rate limited or volume limited? they want different plans.
200 topics polled hourly is 4,800 calls a day, under 4 a minute. Builder covers it at 100/min, 20,000/day.
a 2,000-call backfill wants them at once. 20 minutes at that rate. Scale does it in four.
hot take: most teams build a history table they did not need.
lunarcrush topic spacex
the response hands you the daily average next to the current value. 43,000 mentions in the last 24 hours against a 21,000 average. about 2x, no rows stored.
install is one line on macOS and Linux, and it ships with every paid plan: lunarcrush.com/en/developers…
crypto put up about 521,000 mentions in 24 hours. Bundesliga put up 34,000.
Bundesliga's total reach still came out higher.
that is an 11x gap per mention, and /categories/list/v1 has a row like it for all 41 categories.
every social data vendor revises history. most do it quietly, and the chart you shipped in Q3 changes shape without telling you.
/system/changes logs what can move a historical time series. diff it before debugging your own backfill.
who owns that check on your team?
the endpoint reference, including the system log: lunarcrush.com/en/developers…
people build on LunarCrush at every scale. a script that runs once a week. a bot that pings on every sentiment flip. an agent making calls all day.
a flat monthly plan fit almost none of them. pay-as-you-go does: $1 per 1,000 requests, credits never expire, and errors and 429s don't cost a credit.
same credits across the API, MCP server and CLI. 1,000 free to start: lunarcrush.com/pricing
what's yours: a script, a bot, or an agent?
ask an agent with no live context what is happening with Xbox this week. you get last year, stated with total confidence.
lunarcrush.ai serves the same data as markdown at a path, so it reads instead of guessing.
/topic/xbox this morning: 94.7M engagements in 24 hours.
the markdown endpoints, addressable by topic, creator or category: lunarcrush.com/en/developers…












