ramu11 commented on PR #29454: URL: https://github.com/apache/camel/pull/29454#issuecomment-6737318647
> ### > **[davsclaus](/davsclaus) ** commented [Nov 9, 2026](#issuecomment-6736481107) > • Thanks for pointing this out. I agree that prompt injection risks should be explicitly documented, especially because prompt routes can incorporate content from external systems that may be controlled by untrusted users. I'll add the security guidance to both the AI Prompt component documentation and the MCP Server security section. I'll also clarify the transport Javadocs so they distinguish HTTP support for tools, resources and prompts from stdio support for tools and prompts. Regarding authorization, I agree that `AuthorizationPolicy` would be valuable for prompts, particularly for routes retrieving user- or tenant-specific data. I would prefer to address this in a separate follow-up covering both `ai-prompt` and `ai-resource`, including propagation of the authenticated caller identity into the route exchange. Currently, the HTTP authenticated principal is used in the tool-call path but is not propagated through prompt execution. I'll make sure this limitation is considered in that follow-up. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
