Skip to main content

AI SAST

The AI SAST Analyzer performs static analysis of AI/ML application code, pipelines, and configuration files to detect vulnerabilities, insecure patterns, and coding risks. It scans prompts, agents, and infrastructure-as-code for AI-specific security issues.

Required Blueprints: AWS, GitHub, Bitbucket, Azure DevOps



Sightlines

SightlineDescription
PromptsInsights into AI natural language prompt security across your codebase. Covers system prompts, instruction strings, and prompt templates.
SkillsInsights into LLM tool and skill definition security. Covers tool schemas, function definitions, and agent action groups.
AgentsInsights into AI agent code security and vulnerability posture.
IaC & ConfigInsights into AI infrastructure-as-code and configuration security.

Explorer Node Types

Use these node types in Explorer or KAI to query resources surfaced by this analyzer:

sca.ai.AIResource, sca.ai.Misconfiguration, sca.apibom.PromptDetectionResult, sca.apibom.PromptVulnerability, sca.apibom.SkillDetectionResult, sca.apibom.SkillVulnerability, aws.bedrock.PromptVulnerability


  • AI IAM — Access control for AI models and agents
  • AI DAST — Dynamic testing complements static analysis findings
  • Repository — AI/ML code scanned from source repositories
  • Code — General code security alongside AI-specific findings