Build rich applications with real-time social data. Powered by @LunarCrush.

California, United States
Before you pitch a creator, you want the ten who carried the conversation. lunarcrush creators microsoft --limit 10 --format json 19,470 accounts mentioned Microsoft in the last 24 hours as of this morning, up 18% on the day before. Ranked ten back as JSON you can pipe.
1
1
2,326
social_dominance is a share, so check the denominator before quoting it. Inside technology-brands, last 24 hours as of this morning, YouTube reads 33% and TikTok 26%. That is share of one category. The category is 14.6% of all tracked conversation, which puts YouTube nearer 5%.
198
Feeding an agent social data usually means a scraper, a cleaner, a chunker and a vector store somebody owns. lunarcrush.ai returns the same data as LLM-ready markdown, addressed by topic, creator, category or post. Same bearer key. No pipeline to staff.
1
232
Nobody is asking for another dashboard. They want the numbers inside what they already have open. One key covers the REST API, the CLI and the MCP server. JSON into a warehouse, a notebook, a Slack alert, or an assistant that answers in place. Same header for all of it.
1
207
A volume spike alone is a weak alert. Volume up with sentiment down is the one worth waking someone for. Anthropic on Sept 9: interactions hit 6.3x the trailing 14-day median while sentiment fell from 74% to 24% in a day. /topic/anthropic/time-series/v1. Code in the reply.
1
193
Swap the topic and the 5x threshold to match your own baseline. The last row is a partial day, so it reads r[-2].
21
Your brand monitor watches X. Here is what that leaves on the table. Samsung, last 24 hours as of this morning: 171.7M engagements. YouTube and TikTok carried 91% of them. X carried 8.4%. /topic/samsung/v1 returns the per-network split in the same response.
1
191
Post volume and attention are different quantities. Rank by the wrong one and you collect the wrong rows. /categories/list/v1, last 24 hours as of this morning: Cryptocurrencies sits 9th by mentions, 19th by engagements. 672 engagements per mention. Social Networks: 4,946.
1
229
An average never tells you which row to open. NFL sat at 68% sentiment in the last 24 hours as of this morning. Inside it, the top 15 teams ran from 88% down to 49%. /category/nfl/topics/v1 finds the outlier. /topic/<that>/posts/v1 gives you the posts. Code in the reply.
1
1
5,362
import os, requests h = {"Authorization": "Bearer " + os.environ["LUNARCRUSH_API_KEY"]} b = "lunarcrush.com/api4/public" rows = requests.get(f"{b}/category/nfl/topics/v1", headers=h).json()["data"] worst = min(rows[:15], key=lambda r: r["sentiment"])
17
Before you wire a topic into a collector, check the conversation is the thing you think it is. lunarcrush topic minecraft Last 24 hours as of this morning: 642.3M engagements, 38,896 creators, 65% sentiment. The creators table is the tell. All ten are game channels.
1
268
Your exec digest needs a paragraph up top. Someone keeps maintaining a summarizer for it. /topic/:topic/whatsup/v1 returns a generated summary of a topic's conversation, same bearer header as every call. No prompt to tune, no second model bill. Keep the fields underneath.
1
269
Brand monitoring breaks when your name means something else. HP is also hit points. Apple is also a fruit. Custom searches take inclusion and exclusion terms and return a slug you query like any other topic. Pipe the alerts into Slack or Zapier.
4
255
Prompt for Claude with the LunarCrush MCP connected: where do the OpenAI and Anthropic conversations happen? Two topic calls. OpenAI: 291.3M engagements, 30% of them on TikTok. Anthropic: 138.7M, with 58% on X and 27% on YouTube. Same category, different channel plans.
1
5
5,168