Novembre 2026

November 2026 release notes for Agents in Automation Cloud Dedicated and Test Cloud Dedicated.

November 5, 2026

Garde-fous prêts à l'emploi

You can now protect your agents with out-of-the-box guardrails: predefined safeguards that detect sensitive data, harmful content, and attacks in agent interactions, with no custom configuration or coding. You add them from the Guardrails tab of your agent settings in Studio Web. The following guardrails are available:

• PII detection – Identifies and blocks sensitive information, such as email addresses or physical addresses, in agent prompts, LLM calls, or tool data.
• Harmful content – Detects hate speech, self-harm, sexual content, and violence in LLM interactions, before and after execution.
• Intellectual property protection – Detects intellectual property leakage in model-generated text and code, after execution.
• User prompt attacks – Detects prompt injection and jailbreak attempts before they reach the LLM.

Each guardrail lets you choose what happens when it triggers: log the event, block execution, or send an escalation.

For licensing plans and configuration details, refer to Out-of-the-box guardrails.