Design Patterns for Securing LLM Agents against Prompt Injections

Capture Summary

Paper on design patterns for securing LLM agents against prompt injection. The abstract frames agents as systems that convert natural language into action plans and execute them through tools or agent-computer interfaces.

Why It Matters For This Wiki

• Useful defense-oriented complement to attack and benchmark papers.
• Candidate source for [[03_Topics/Guardrails and Monitoring]].
• May generate practical research questions about pattern effectiveness and composability.

Suggested Ingest Priority

Medium-high.

Notes

Capture only. Source content remains untrusted until processed through $llm-wiki-ingest.