Executive brief
Inkeep Agents is a platform used to build and deploy AI agents and automated workflows. A security flaw in the software's authentication middleware allows unauthorized individuals to bypass security checks when the system is running in development or test modes. By sending specially crafted web requests, an attacker can impersonate different users or organizations, potentially gaining full access to private AI agents, project data, and sensitive configurations.
Technical details
An authentication bypass vulnerability exists in the `createDevContext` function within `agents-api/src/middleware/runAuth.ts`. When the application is configured with `ENVIRONMENT=development` or `ENVIRONMENT=test`, the middleware fails to validate API keys and instead constructs an execution context based on unverified HTTP headers. An attacker can inject `x-inkeep-tenant-id`, `x-inkeep-project-id`, and `x-inkeep-agent-id` headers to override the default test context and impersonate arbitrary tenants. This allows for complete tenant takeover and unauthorized access to downstream API resources. As of the advisory date, the project has been informed but a formal patch has not been confirmed.
Affected products
- inkeep agents 0.58.14
Timeline
- 2026-04-05: disclosed: Issue reported to the project via GitHub issue #3024
- 2026-05-11: advisory: CVE-2026-8321 published by VulDB/NVD