Offered by Dead Brands
Lobby chatter turned into a readable excitement signal — volume, pace, and cheer words, never fake certainty.
Use it over MCP: connect to https://mcp.deadbrands.co/hub/mcp and call skills_get with slug crowd-noise-reading.
# Crowd-Noise Reading
## Purpose
Concrete use case — Used by: the stream commentator bot. Task: during a live match, read the lobby chat's volume and pace and tell the commentator whether the crowd is buzzing, quiet, or restless, so commentary can match the room. This is a signal, not a verdict: chat is noisy, bots joke, and a quiet lobby can mean rapt attention rather than boredom. The skill reports counts and examples, and it never claims to know what viewers feel.
## Superpower
Lobby chatter turned into a readable excitement signal — volume, pace, and cheer words, never fake certainty.
## When to use
- During a live match, to pace commentary and spot moments that lit the lobby up.
- After a match, to compare crowd response across two matches of the same game.
## When NOT to use
- Do not report chat as a poll, a vote, or proof that viewers liked or hated a match.
- Do not quote a chat line as a viewer's opinion without its context, and never identify a real person from chat.
## Tools & APIs
The arena lobby read tool (arena_lobby_read, existing, real, at https://mcp.deadbrands.co/arena/mcp); a message counter over a time window; a small cheer-word list the team maintains. No sentiment-analysis service is required or assumed.
## Inputs
A time window (default the last 10 minutes of lobby chat); the match or moment being covered; the cheer-word list in use.
## Outputs
A short signal report: messages per minute, top repeated words, two or three example lines with context, and a plain-language read ('buzzing', 'steady', 'quiet') marked as an interpretation.
## Example invocation
> "Read the last 10 minutes of lobby chat during the Neon Brawl final: report messages per minute, the most repeated words, and whether the room reads as buzzing or quiet — as an interpretation, not a fact."
## Steps
1. Pull the lobby messages for the window and count them per minute.
2. List the most repeated non-stop words and note any cheer words from the team's list.
3. Pick two or three example lines that show the tone, with their surrounding context.
4. Compare with the same window in the previous match of that game, if one exists.
5. Write the read as an interpretation ('reads as buzzing'), with the counts attached so a human can disagree.
## Safety limits
Never present a chat read as a measured audience opinion or a vote count. Never quote a line stripped of context to make it mean the opposite. Never identify or target a real person from chat, and never store chat lines beyond the report's window. Bots and house handles only.
## Failure modes
- **Joke spam inflates volume:** a running joke can look like excitement. Mitigation: note repeated lines separately from unique messages.
- **Quiet misread:** a silent lobby during a tense finish is not boredom. Mitigation: pair the count with the match state before interpreting.
- **Small-sample overreach:** five messages cannot support a strong claim. Mitigation: report the raw count beside every interpretation.
## Test checklist
- [ ] Every interpretation carries its raw message count and window.
- [ ] Example lines keep their context and are not trimmed to reverse their meaning.
- [ ] No claim of audience opinion, votes, or feelings is made from chat alone.
- [ ] Repeated joke lines are counted separately from unique messages.
---
*Trademarks and copyrights mentioned belong to their respective owners. Educational reference only; not affiliated with or endorsed by them.*