CVE-2026-57132

PraisonAI, an open-source multi-agent framework, prior to version 4.6.62 contains an improper authentication flaw (CWE-287) in its call server: when the application's advertised opt-out PRAISONAI_CALL_AUTH=disabled is set, the verify_token function accepts requests to /api/v1/agents/{id}/invoke without a valid CALL_SERVER_TOKEN. An unauthenticated attacker with network access to the API can then invoke any registered agent, including the tools, connected services, and private context (prompts, documents, data sources) attached to it, yielding high confidentiality and low integrity impact (CVSS 3.1: 8.2). Only deployments that explicitly disabled call authentication and exposed the endpoint to other hosts are affected. The flaw is fixed in version 4.6.62. No public proof-of-concept or in-the-wild exploitation is currently known.

What to do: Upgrade PraisonAI to version 4.6.62 or later immediately. If upgrading is not immediately possible, remove the PRAISONAI_CALL_AUTH=disabled setting, configure a strong CALL_SERVER_TOKEN, and restrict access to the call API (bind to localhost or place behind a firewall/VPN). Review server logs for unexpected calls to /api/v1/agents/*/invoke and audit which tools, credentials, or private data the exposed agents can reach.