Zambo guide
Test MCP Prompt Injection
Paste any suspicious text into Zambo’s prompt_shield tool and get a check back with a receipt, before that text ever reaches your agent.
Add Zambo to your AIWhat prompt injection looks like
Start with one small, low-risk request that matches test MCP prompt injection. Review the returned result, open its receipt and note anything the record cannot establish.
Use the connection guide to connect an AI app first.
Test a sample now
Apply test MCP prompt injection to one concrete task. Keep the requested action, the returned result and the material used to review it together, then label anything that was not checked.
What the check catches
Before relying on test MCP prompt injection, compare the actual result with the original request and follow any linked source. A recorded action can aid review, but it does not establish that every conclusion is correct.
Add the shield to your agent
Paste any suspicious text into Zambo’s prompt_shield tool and get a check back with a receipt, before that text ever reaches your agent.
Start by answering the specific question “test MCP prompt injection” with information you can inspect. Separate observed facts from assumptions and unresolved points.
Try this prompt
Start with a small request. Change the details to match your task.
Use Zambo to help with test MCP prompt injection. Show the steps you actually ran and give me a link I can review.
FAQ
What is test MCP prompt injection?
Paste any suspicious text into Zambo’s prompt_shield tool and get a check back with a receipt, before that text ever reaches your agent.
What does a Zambo receipt show?
It shows the saved record and its available status. It does not prove that the output is correct or complete.
How do I try this with Zambo?
Add Zambo to a supported AI app from /add, ask it to use Zambo for the task, then open the returned receipt link.
Try it with your AI
Connect Zambo once, then ask your AI to use it for a real task.
Add Zambo to your AI Try Zambo Chat