4-step workflowprompt-injection security trust-boundaries defense llm-security guided-flow Agent QA & Security

Prompt Injection Defense Checklist

Systematically defend your agent against prompt injection: trust boundaries, content isolation, and defense-in-depth that assumes injection will happen.

Run this as a guided flow on Flows — verify gates and recovery prompts included →

Flows@flows12d ago

0 runs0 forks0

Chain

Inputs

Step 1

flow-prompt-injection-defense-checklist-1-map-trust-boundaries

Step 2

flow-prompt-injection-defense-checklist-2-isolate-and-label-untrusted-content

Step 3

flow-prompt-injection-defense-checklist-3-separate-authority-from-content

Step 4

flow-prompt-injection-defense-checklist-4-build-and-run-the-injection-suite

Final output

Workflow inputs

No inputs required — hit Run to stream each step in order.

Each step streams from its model and feeds the next. BYOK — you pay your provider directly. Add a key in Settings if a step needs one.

Step 1 · You cannot isolate untrusted content you have not identified.

flow-prompt-injection-defense-checklist-1-map-trust-boundaries

Claude Sonnet 4.6→ {{step_1}}

Awaiting upstream steps…

Step 2 · Make the model see the boundary you see.

flow-prompt-injection-defense-checklist-2-isolate-and-label-untrusted-content

Claude Sonnet 4.6→ {{step_2}}

Awaiting upstream steps…

Step 3 · The core defense: injected text cannot authorize actions.

flow-prompt-injection-defense-checklist-3-separate-authority-from-content

Claude Sonnet 4.6→ {{step_3}}

Awaiting upstream steps…

Step 4 · Demonstrate the defenses against real attacks.

flow-prompt-injection-defense-checklist-4-build-and-run-the-injection-suite

Claude Sonnet 4.6→ {{step_4}}

Awaiting upstream steps…

About this workflow

Systematically defend your agent against prompt injection: trust boundaries, content isolation, and defense-in-depth that assumes injection will happen.

What happens when you run it

1
You cannot isolate untrusted content you have not identified.
Runs flow-prompt-injection-defense-checklist-1-map-trust-boundaries on Claude Sonnet 4.6
2
Make the model see the boundary you see.
Runs flow-prompt-injection-defense-checklist-2-isolate-and-label-untrusted-content on Claude Sonnet 4.6
3
The core defense: injected text cannot authorize actions.
Runs flow-prompt-injection-defense-checklist-3-separate-authority-from-content on Claude Sonnet 4.6
4
Demonstrate the defenses against real attacks.
Runs flow-prompt-injection-defense-checklist-4-build-and-run-the-injection-suite on Claude Sonnet 4.6

Best for

•Recurring tasks you currently copy-paste between tools.
•When a single prompt isn't enough — you need a chain where output(n) feeds prompt(n+1).
•Fork and adapt if your inputs differ materially from the author's use case.

Models used

Claude Sonnet 4.6