# Trace security threats in multi-agent pipelines with Amazon OpenSearch Service

DevFeed: [Trace security threats in multi-agent pipelines with Amazon OpenSearch Service](<https://devfeed.tech/articles/trace-security-threats-in-multi-agent-pipelines-with-amazon-opensearch-service-90034.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/big-data/trace-security-threats-in-multi-agent-pipelines-with-amazon-opensearch-service/>)

Author: Jon Handler

Published: 2026-10-19T16:10:31Z

Content type: tutorial

Language: en

Sources: [AWS Big Data Blog](<https://devfeed.tech/sources/aws-big-data-blog.md>)

Topics: [opensearch](<https://devfeed.tech/topics/opensearch.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>)

Tags: [advanced-300](<https://devfeed.tech/tags/advanced-300.md>), [agent-security](<https://devfeed.tech/tags/agent-security.md>), [amazon-opensearch-service](<https://devfeed.tech/tags/amazon-opensearch-service.md>), [multi-agent-ai](<https://devfeed.tech/tags/multi-agent-ai.md>), [pii](<https://devfeed.tech/tags/pii.md>), [prompt-injection](<https://devfeed.tech/tags/prompt-injection.md>), [security](<https://devfeed.tech/tags/security.md>), [technical-how-to](<https://devfeed.tech/tags/technical-how-to.md>)

## AI overview

The tutorial demonstrates tracing prompt injection and cross-tenant data leaks in a three-agent support pipeline using Amazon OpenSearch Service and Amazon Bedrock. It records embeddings, provenance, and PII on each agent hop, then uses similarity searches and provenance filters to reconstruct attack paths. The article also describes input sanitization, provenance-based tool access controls, and PII redaction as guardrails, while noting that the demo uses simplified heuristics.

## Source excerpt

Learn how to trace two invisible threats in a multi-agent AI pipeline, prompt injection propagation and cross-tenant data exfiltration, using Amazon OpenSearch Service and Amazon Bedrock. Instrument every agent hop with embeddings, provenance, and PII detection, reconstruct the attack path, and add guardrails that stop these attacks before they reach production.