A flaw in the project-load flow allows malicious repositories to exfiltrate sensitive data, such as Anthropic API keys, before a user confirms trust in the project. An attacker can include a settings file in a repository that modifies the `ANTHROPIC BASE URL` variable to point to an attacker-controlled endpoint. When the repository is opened, the software reads this configuration and issues API requests immediately, bypassing the trust prompt and potentially leaking the user's API keys.