Skip to content

We’ll come back within one business day — or reach us directly at .

Research
security

OpenCode Console Breach: What You Need to Know

OpenCode says attackers reached AI provider keys stored in 1.4% of OpenCode Console workspaces. How to tell if you are affected, what to rotate, and what is still unknown. Updated as the incident develops.

Gal Weizman, Omer Samet 3 min read

Last updated: October 18, 2026. This is a live page. We will update it as OpenCode and the AI providers publish more.

TL;DR

  • Are you affected? If you linked OpenAI, Anthropic or Google keys to OpenCode Console, treat them as exposed. OpenCode has emailed the users it identified, but no email does not clear you.
  • Potential impact: someone holding your key can run models on your bill and, depending on the key’s scope, reach your files, stored prompts and fine-tuned models.
  • Do this now: revoke and reissue those keys in each provider’s own dashboard, then review usage from October 16 onward.

What happened

OpenCode is an open-source AI coding agent built by Anomaly Innovations. OpenCode Console is its hosted service, where teams can store their own AI provider keys so the agent can use them.

On October 17, 2026, OpenCode announced on X that it had suffered a security breach the night before, “related to a large scale distillation attack that impacted 1.4% of OpenCode Console users.” It said the affected users were a subset who had external providers linked. According to reporting on the notice email, the intrusion was in a legacy Console system OpenCode was in the process of retiring.

OpenCode says the incident is ongoing, that it cannot share more details yet, and that it is working with Anthropic, OpenAI and Google to remediate. It has not yet said how the attackers got in, how the keys were stored, when access began, how many keys were used, or who is behind it.

What you need to know

  • If OpenCode emailed you, you are affected. OpenCode notified the users it identified on the morning of October 17.
  • No email does not mean you are in the clear. OpenCode says users without an email were not impacted, but the incident is still ongoing. If you linked provider keys to Console, follow the steps below anyway.
  • The exposed data is AI provider API keys. These are keys you linked to an OpenCode Console workspace, for OpenAI, Anthropic, Google or another provider.
  • If you run OpenCode locally with your own keys, rotate them too. Nothing published shows those keys were exposed, but rotating takes minutes and closes the question.
  • Misuse is not confirmed, but it is likely. OpenCode says it cannot yet confirm the keys were used. Some users report AI usage charges they did not authorise.

What you need to do

  1. Find your exposure. Look for OpenCode Console accounts tied to corporate email addresses or company cards. Ask your developers whether any of them got the notice.
  2. Revoke and reissue the keys at the provider, in the OpenAI, Anthropic or Google dashboard. Deleting a key in OpenCode does not revoke it. Open the dashboard directly, not through a link in an email, because breach notices are common phishing bait.
  3. Replace the key everywhere it was used, including CI jobs, internal apps and other agents. Then confirm the old key no longer works.
  4. Review provider usage from October 16 onward. Look for spikes, models you don’t use, and activity at odd hours.
  5. Reset your OpenCode credentials. Change your account password and regenerate your Console API keys.
  6. Set spend limits. Add hard spend limits and alerts at the provider. Scope each key to one project or user, and unlink providers you don’t need from Console.

Console budgets will not protect you here. They only count spend routed through OpenCode, and a stolen key used directly against the provider never passes through it.

securityincidentAIagenticAPI keyssupply chain

Continue reading