A security breach affecting open-source AI agent framework provider OpenCode could expose API keys linked to users of Anthropic, OpenAI and Google AI models.
On October 17, OpenCode confirmed a security breach that occurred the night before and affected its Console service, which allows teams to share access to AI models through centrally managed API credentials.
Rather than exposing the underlying OpenAI, Anthropic or Google keys to individual team members, Console stores each workspace’s credentials and uses them to authenticate requests to the respective providers.
OpenCode is a widely used AI agent platform designed for software developers. It has 217 stars on GitHub.
In a July 2026 interview with Y Combinator – financial backer of Anomaly, the company which develops and maintains OpenCode – the founder and CEO Jay V claimed OpenCode had reached approximately 13 million monthly active users and 4.6 million weekly active users.
OpenCode Points to Unauthorized Access to Legacy System
The incident was “related to a large-scale distillation attack that impacted 1.4% of OpenCode Console users,” said OpenCode on social media.
“We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked,” the company added.
The email was made public by several OpenCode users on the dedicated subreddit forum and the official Discord channel.
The email said OpenCode had "identified unauthorized access to a legacy system that we were in the process of decommissioning" and that "third party provider keys (e.g. OpenAI, Anthropic) saved in your workspace may have been exposed."
Specifically, Dax Raad, one of the main contributors to OpenCode and co-founder of Anomaly, noted on X that the breach involved only "the OpenCode Console, not OpenCode itself.”
OpenCode also notified Anthropic, OpenAI and Google to remediate the issue.
Recommended Steps to Mitigate OpenCode Threat
In an incident response report, software supply chain security firm Safeguard provided recommendations for OpenCode users to identify whether they have been affected by the incident and mitigate its impact:
Key steps include:
- Checking whether you may be affected: search your inbox for the subject line “Action Required: Potential provider key breach.” If you have received the email, follow the guidance provided. If not, still review your keys as a precaution
- Identifying all potentially exposed keys: list every OpenAI, Anthropic and Google Gemini API key that was ever stored in an OpenCode Console workspace, including keys added by other workspace admins
- Reviewing API usage for suspicious activity: check each provider’s usage and billing reports from 16 October onward, and further back if possible. Look for unexpected usage spikes, unfamiliar models or unexplained charges
- Creating replacement keys: generate new API keys for any potentially exposed credentials before revoking the old ones. Use project- or workspace-specific keys with appropriate restrictions and spending limits
- Revoking potentially exposed keys: disable or delete the old keys through the relevant provider dashboard. Do not assume a key is safe simply because it has not shown suspicious usage
- Updating all systems using the old keys: replace them in CI/CD secrets, environment files, local agent configurations, production applications, and team members’ machines. Ensure all consumers use the new keys before removing any remaining dependencies on the old ones
- Verifying that old keys no longer function: test revoked keys against the relevant provider APIs and confirm they return an authentication error rather than a successful response
- Applying additional security controls: restrict keys to the required projects and APIs, set appropriate spending limits and budget alerts, and review workspace access and permissions
- Securing your OpenCode Console configuration: if you continue using Console with your own provider credentials, add a dedicated, restricted API key rather than reusing a production key.
OpenCode and Anomaly have been contacted with a request for comment.










