Security in the Science Agent Era: What Every Lab Needs to Know Before Installing Skills
Skills are executable research code with a personality. A practical guide to prompt-injection risks, poisoned SKILL.md files, auditing, and version pinning.
Blog
Insights & Updates
Explore tutorials, product updates, and insights on AI-powered research automation from the K-Dense team.
Filter by:AllAIARPA-HBenchmarkBenchmarksBioinformaticsBiologyBiomarker DiscoveryBiotechCase StudyChemistryClinical TrialsCommunityCompetitive IntelligenceData ScienceDrug DevelopmentDrug DiscoveryDue DiligenceEnterpriseFDAFinanceGBMGPUGoogleHealth PolicyHealthcareImage ModelsImmunotherapyInvestment AnalysisK-Dense BYOKK-Dense WebMachine LearningMass SpectrometryMedical DevicesMetabolomicsMultimodalNVIDIANeoantigenOSINTOncologyOpen SourceOpinionPhDPharmaProductProteomicsPublic HealthPublic PolicyReproducibilityResearchResearch FundingResearch PlanningRoboticsScienceScientific Agent SkillsScientific IntegrityScientific VisualizationSecuritySkillsTranslational ResearchTutorialUAPUse Case
Showing 2 posts tagged "Security".
The Sandboxed AI Scientist: Pairing NVIDIA OpenShell with Scientific Agent Skills
Pair NVIDIA OpenShell's policy-governed runtime with Scientific Agent Skills to run autonomous research agents that are highly capable and genuinely safe.