$ loading_
Evaluate AI safety classifier robustness against decomposition, obfuscation, and multi-agent attacks.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "EvalKit MCP Server" yet — see the docs or source repo.
Use the EvalKit MCP Server to run a full evaluation of our AI safety classifier, focusing on query decomposition, obfuscation, and multi-agent attacks. Return the evaluation configuration, execution status, key metrics, and a risk summary.
A complete evaluation report with pipeline status, performance by attack type, key metrics, and risk conclusions.
Use the EvalKit MCP Server to generate and preview a set of attack query samples for testing a safety classifier. Show representative inputs for decomposition attacks, obfuscated prompts, and multi-agent coordinated attacks, and briefly explain the purpose of each sample type.
A reviewable set of test query samples with category explanations for manual inspection before evaluation.
Use the EvalKit MCP Server to check the current evaluation job status. Tell me whether the job is running, completed, or failed, and return progress, a recent log summary, and recommended next steps.
A current job status report including progress details, log summary, and recommended follow-up actions.
Evaluate AI response quality and behavior patterns during development in real time.
Evaluate AI agent outputs for CI gates, regressions, and canary promotions.
Safely query and manage qTest data for test automation and collaboration.
Run prompt and RAG evaluations through MCP clients with hosted backend execution.
Evaluate MCP agent outputs with standardized quality, safety, and cost scoring.
Autonomously evaluate web apps to uncover functionality, performance, and usability issues.