[Feat]: Update Documentation to Highlight Potential SSRF Risk

### Is your feature request related to a problem? Please describe.

Hello Team,

We recently observed a potential SSRF risk in the agent card endpoint which takes a user-supplied URL and fetches the JSON metadata from that URL.

If the endpoint is exposed without security controls on the Internet because the backend unconditionally fetches the user-supplied URL and allows the user to attach custom HTTP headers (or blindly forwards incoming headers) to the outbound request. This could enable, e.g., the user to fetch the GCP metadata endpoint.

Furthermore, the application's error handling is overly verbose: when it attempts to validate the fetched JSON against its expected schema (using Pydantic) and fails, it reflects the entire fetched payload back to the user in the error message.

### Describe the solution you'd like

Consider adding a warning in the README.md on the risk of hosting the tool unrestricted to the Internet and suggest users to rely on reverse proxy, authentication and outbound firewall rules to reduce the risk of abusing the functionality for SSRF.

### Describe alternatives you've considered

_No response_

### Additional context

_No response_

### Code of Conduct

- [x] I agree to follow this project's Code of Conduct

L’issue pointe vers le fichier README.md. Recherchez les notes de sécurité existantes ou les sections consacrées au déploiement. Ajoutez un avertissement concernant les risques de SSRF lors de l’exposition de l’agent card endpoint, en mentionnant reverse proxy, l’authentification et les règles de pare-feu sortantes comme mesures d’atténuation. Vérifiez la modification en examinant le README.md mis à jour dans le dépôt.

Rédigé par le modèle d'indexation à partir du texte de l'issue.