Twitter Bot & Coordination Detector

An analysis skill that scans the engagement network forming around an X/Twitter account (the actors who reply, quote, like, retweet) and classifies each actor as organic / amplifier / suspected-bot / coordinated. Given a target handle or list of tweets, it pulls the profile + historical data of the actors who engaged with those tweets; across 6 signal axes (account age × engagement anomaly, timing bursts, copy-paste template replies, follower/following ratio, generic-phrase density, cluster coherence) it assigns each actor an anomaly score between 0 and 1.

When is it used: to figure out how much of the engagement behind a viral tweet is real (viral autopsy), fake follower/engagement audits, hype-network mapping, coordinated inauthentic behavior (CIB) detection.

Critical rule: Every claim is backed by at least 2 independent signals; nothing is called a "bot" on a single signal. Classification is probabilistic (low/medium/high/conclusive evidence strength); the final judgment is left to human interpretation.

Output: A Markdown report (with a JSON core) containing the classification distribution, per-actor axis scores + an evidence summary, the detected coordination clusters, and an evidence-strength rubric. It uses only public engagement data; it does not analyze DMs or private metadata.