GitBook’s status page flagged a service outage and ongoing security remediation after the AI‑agent breach.
*GitBook suffered a targeted AI‑agent attack on Nov 2, 19:00 CET. The company shut down its app and API, replaced every credential, and warned customers to rotate tokens. No evidence yet of content tampering, but exposure of secrets is confirmed.*
GitBook disclosed a security incident that began at 19:00 CET on November 2. An attacker using sophisticated AI agents breached infrastructure credentials, prompting the platform to pull its app, API, and authenticated documentation sites offline. The breach was detected instantly, and GitBook moved to block further impact while keeping publicly cached docs visible. No official breach notice has reached customers beyond the support ticket that sparked this report, leaving users scrambling for guidance.
At 19:00 CET on November 2, GitBook’s monitoring flagged unauthorized access to infrastructure credentials. The intruder employed AI agents in a highly targeted attack, compromising internal secrets. GitBook’s security team identified the breach immediately and isolated the affected components. To prevent further damage, the company took the GitBook app, API, and authenticated documentation sites offline, while serving cached public content. The rapid detection limited the window of exposure, but the incident forced a full service outage for several hours.
GitBook’s response was swift and comprehensive. The attacker was blocked and the vulnerabilities that allowed entry were closed. Every credential and secret across the infrastructure was replaced. Services were only restored after confidence in their security was re‑established. All personal API tokens were revoked, and every customer’s API tokens were invalidated. The company also issued a directive to rotate any third‑party API tokens or secrets used to connect external tools, and to re‑authenticate any GitBook MCP servers.
GitBook’s investigation identified four categories of potentially accessed data: Git Sync secrets, GitBook API tokens, third‑party API tokens used for external integrations, and questions asked to the GitBook Assistant by site visitors. The company found no evidence that any user‑generated content within GitBook was altered. However, the exposure of authentication tokens and integration secrets poses a risk of downstream attacks on connected services.
GitBook urges every organization to rotate existing API tokens immediately, as all personal tokens were revoked during the incident. Users must also rotate any third‑party API tokens or secrets tied to external tools, and update any published integrations that embed secrets. Those operating a GitBook MCP server need to re‑authenticate it. The guidance advises contacting the organization’s GitBook admin for assistance and monitoring the status page for further updates.
The breach began at 19:00 CET on November 2, when an attacker gained unauthorized access to infrastructure credentials.
GitBook says the attacker may have accessed Git Sync secrets, GitBook API tokens, third‑party API tokens, and questions asked to the GitBook Assistant.
Users must rotate all GitBook API tokens, rotate any third‑party API tokens or secrets, and re‑authenticate any GitBook MCP server.
GitBook is now monitoring for additional threats and promises new security measures. A full post‑mortem will be published soon, detailing the breach timeline and remediation steps. Until then, users face the burden of sweeping credential rotations and heightened vigilance over connected services. The incident underscores the growing risk of AI‑augmented attacks on SaaS platforms and the need for rapid, transparent response protocols.
Sources: [GitBook security incident update posted on Hacker News]