French vendor pairing an open-source Python testing library with a delivered assessment service. Automated and expert-led testing probes conversational agents for prompt injection, data disclosure, sycophancy, hallucination and inappropriate refusals, returning a severity-ranked vulnerability report and a signed go/no-go deployment recommendation.
hybrid · generally available · Research snapshot 2026-09-06
Show me a full assessment report for my customer-facing agent, with vulnerabilities ranked by severity and the go/no-go deployment recommendation.
Capabilities and evidence
Support labels reflect the supplied research. Documentation and vendor claims are not independent product tests. “Not established” means the researcher did not find support; it does not prove a capability is absent.
Documented by provider
Giskard Open Source is a Python library whose LLM Scan combines heuristic and LLM-assisted detectors to generate synthetic and adversarial test cases for issues including prompt injection and stereotypes, while RAGET generates question, reference answer and reference context sets to test RAG agents.
Limit: The library documentation does not describe tracing, access control, retention or audit records.
The company positions Giskard Hub as AI security delivered as a service, producing a full vulnerability report ranked by severity, remediation guidance and a go/no-go deployment recommendation signed by its AI security experts.
Limit: Product page; the specific report contents, tester qualifications and cadence are not documented in reference material.
Documented test categories include sycophancy, prompt injection, data disclosure, inappropriate content, inappropriate denials, contradictions, hallucinations and omissions.
Limit: Category list appears on the marketing page without documented detection methodology per category.