CommTox, a community perceived toxicity classifier, leverages machine learning and historical behavioral data in order to assign a score for the perceived toxicity within a community. Building contextual awareness on top of Perspective API, CommTox brings highly flexible and adaptable context-aware toxicity classification to developers. CommTox is deployed across YouTube, and evaluates perceived toxicity across a number of online communities. Perceived toxicity is community-specific and incorporates historical community behavior into the decision making. Results indicate that while community reception of comments offers a fair amount of accuracy when predicting perceived toxicity, the most effective models need to include some level of text-dependent features, such as word-embeddings. The goal of CommTox is to provide a standardized metric from which comparisons free of bias can be drawn for varying communities.