Source record

Published by Practical AI (podcast) (T1.5 source). The original is at https://share.transistor.fm/s/e60866e6.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

Podcast episode on security risks in agentic AI systems. Practitioner Donato Capitella discusses real-world penetration test findings, design patterns for LLM agents, and his open-source red-teaming toolkit Spikee (Simple Prompt Injection Kit for Evaluation and Exploitation).

Hands-on practitioner perspective on agent security, with a concrete open-source tool for testing prompt injection and data exfiltration in LLM agent pipelines.

Cited by

No citations on record.