Chpokai
Authorized AI-agent security, pentesting and acceptance testing platform.
CHPOKAI executes controlled, reproducible behavioral security assessments against web systems, APIs and AI agents. It identifies prompt injections, business logic bypasses, auth leakage, and API flaws, generating auditable proof-of-exploit reports.
Autonomous adversarial verification with machine-readable evidence.
Engineered with production-grade isolation, deterministic reliability, and seamless integration across the XIK infrastructure ecosystem.
Core Capabilities
4 Functional ModulesAutonomous Attack Simulation
Generates non-destructive exploitation vectors testing for OWASP Top 10 for LLMs and web APIs.
Sandbox Execution Proofs
Isolates findings with step-by-step reproduction traces and verifiable network payloads.
Guardrail Regression Testing
Continuously asserts that system prompts and safety filters maintain their integrity across model updates.
Machine-Readable Audit Reports
Exports formal vulnerability logs in JSON, Markdown, and SARIF formats for CI/CD integration.
Technical Specifications
- Identifier
- chpokai
- System Stack
- Python Security Framework, Sandboxed Headless Chromium, Multi-LLM Swarm, Next.js UI
- Runtime Architecture
- Isolated Ephemeral Containers
- Deployment Model
- Secure VPC
- Supported Interfaces
- HTTP/2WebSocketCLI EngineCI/CD Action