POCFORGE / SECURITY
AI / LLM Security Testing India
Security testing for AI and LLM applications, RAG systems and agentic workflows, focusing on prompt injection, sensitive-data exposure and unsafe tool execution.
What we test
- Direct and indirect prompt injection
- System prompt and instruction leakage
- RAG data isolation and retrieval abuse
- Tool and function-call authorization
- Cross-user and cross-tenant data exposure
- Agentic attack chains and unsafe actions
Assessment outcome
We demonstrate practical attack paths and provide remediation guidance for model, application and authorization controls.