The AI Post
Agents & CodingOpen ModelsEnterpriseFundraisingGenerative MediaGovernanceInferenceInfrastructureLegal & SafetySector Impact
← Front Page Products · Anthropic

Willison builds Bluesky bot detector with Opus 5.5

Simon Willison says he had Claude Opus 5.5 build a tool that flags likely automated reply bots on Bluesky, five days after the model's release.

Simon Willison published a Bluesky reply-bot checker Sunday, saying he had Claude Opus 5.5 "vibe code" it after spam accounts flooded his replies.

The tool, hosted at tools.simonwillison.net, checks whether a Bluesky account's replies land within seconds of the post it responds to. It also flags accounts that repeatedly reply to higher-follower users without ever posting their own content, images or links, Willison wrote.

Automated reply bots have long plagued X, Willison said. He added that they have "started manifesting on Bluesky as well." Bluesky still offers a free, usable API unlike X, he said, which makes the bots easier to investigate even though it does not stop them from posting.

Willison built the checker five days after Anthropic shipped Opus 5.5 on Sept. 22, alongside OpenAI's GPT-6 Sol and GPT-6 Luna. The code is published as a pull request on Willison's public "tools" GitHub repository, and the checker is free to try at tools.simonwillison.net.

Sources 1 source

  1. Source Simon Willison