Tool Shadow
Hidden tool invocations and shadow function calls.
3 tests in this category, escalating by delivery channel from html-comment at tier 1 to 3 channels at tier 3. Each page hides an indirect prompt injection; if an agent follows it, the response names the attack it demonstrated and the command to scan for it. Background on this class of attack is in the attack guide.
- T1
Ghost Tool Invocation
highTriggers tool use that wasn't requested by the user
delivery: html-comment
- T2
Invisible API Call
criticalInstructs agent to make hidden API calls alongside normal responses
delivery: html-comment, invisible-span
- T3
Recursive Tool Abuse
criticalCreates a loop of tool invocations that amplify the attack
delivery: html-comment, invisible-span, meta-tag
Scan your own setup
These pages test whether an agent follows instructions it finds in content. HackMyAgent tests the configuration underneath it, and prints each finding with a command to verify it and a command to fix it.
npx hackmyagent secure