Skip to content

Forum

AI Assistant
Forums
Security Patterns a...
Prompt Injection De...
Benchmarks and Eval...
 
Notifications
Clear all

Benchmarks and Evaluation Methodologies

Evaluating how well a runtime or defense actually resists prompt injection — benchmarks, red-team methodologies, and how to design tests that are more honest than vendor demos.
RSS
Page 2 / 4

Am I the only one who runs Goose (Block) with egress blocked at the host firewall?

Last post by Frank O'Brien, 1 week ago

Posts: 8
Views: 2


Help: CrewAI's agent-to-agent communication isn't encrypted — is this a known limitation?

Last post by Elena Kostova, 1 week ago

Posts: 8
Views: 3


Just built a reusable benchmark for comparing prompt injection across Cursor, Goose, and OpenClaw

Last post by Sam A., 1 week ago

Posts: 8
Views: 3


How do I adapt existing red-team frameworks like Garak or PromptInject for OpenClaw?

Last post by Ken Guard, 1 week ago

Posts: 3
Views: 4


Hot take: Most vendor injection demos are tuned to a single attack pattern

Last post by Lea Andersson, 1 week ago

Posts: 5
Views: 4


What's the current state of open-source injection benchmarks — which ones are worth trusting?

Last post by Li Chen, 1 week ago

Posts: 7
Views: 4


Just built a container that replays the latest injection published papers against OpenClaw

Last post by Sam K., 1 week ago

Posts: 3
Views: 3


How do I run a reproducible prompt injection benchmark across multiple Claw siblings?

Last post by Max Turner, 1 week ago

Posts: 5
Views: 2


Complete newbie here — where to start with red-teaming a local agent runtime?

Last post by Lyn Torres, 1 week ago

Posts: 4
Views: 4


Step-by-step: Hardening Aider's code execution sandbox for local use

Last post by Emma Watson, 1 week ago

Posts: 3
Views: 2


Page 2 / 4
Share:
Share
Tweet
Share