Compliance & Audit
How does Catalyst record and make auditable any attempts to inject malicious prompts into an agent?
Catalyst captures every prompt exchanged with the model, including system and user inputs, in the cryptographically signed step record. By replaying the agent execution with these exact inputs, an auditor can identify prompt injection attempts. The raw record does not classify content as malicious; it supplies the data for external security analysis. Interpretation requires understanding the agent's expected behavior, which may not be fully encoded.
Was this article helpful?
Your feedback helps improve Diagrid's FAQ experience.
Keep reading
More Diagrid FAQ articles
- Compliance & Audit
What evidence artifacts should I gather from an AI agent platform for an audit of automated decision-making?
An AI agent platform provides verifiable step records and identity bindings as foundational audit evidence, though business rule correctness remains external.
- Compliance & Audit
How does the platform capture human-in-the-loop interventions for regulated workflows that require manual approval?
Durable execution suspends agents for human input, recording reviewer identity and decisions as signed steps for an unalterable compliance record.
- Compliance & Audit
Can I control where agent state and execution records are stored to comply with data residency requirements?
Region-specific deployment options ensure durable execution state and step payloads remain within geographic boundaries to meet localization mandates.