Your agent hit the context limit. Here's the playbook

An agent humming along can stop mid-task and declare the job done when it isn’t. One common cause is the AI context limit, a hard ceiling on how many tokens a model can hold in its working memory at once. This playbook covers how to recover.