UI & Experience
VenomX surfaces through a customized OpenWebUI frontend: a clean, security-focused chat interface with purpose-built components for target management, formatted vulnerability reports, and structured output.
Interface
Natural language in. Structured security intelligence out. The chat interface connects directly to the VenomX backend. No cloud relay, no external API, fully on your network.
Feature Set
Four purpose-built capabilities beyond standard chat, each designed around real security workflows.
The primary interaction layer. Users submit natural language queries (scoping questions, exploit lookups, CVE triage requests, report generation) and VenomX responds with structured, formatted output. Powered by OpenWebUI's conversation management with custom system prompt injection to keep responses security-domain specific.
Capabilities
Scan results don't just appear as static output. They are organized into a structured target management workflow that centralizes hosts, asset details, and scan status in one place. As the agent completes discovery and assessment, the system updates target records in real time so it is easier to track scope, compare findings, and prioritize what to review next.
Management Capabilities
VenomX formats its output as machine-readable security reports, not free-form text. The renderer handles CVSS score tables, MITRE ATT&CK technique references, remediation steps, and tool command output in consistent, copyable formats. Reports are generated directly in the chat window and can be exported as Markdown.
Report Sections
OpenWebUI's default theme is replaced with VenomX's design system: dark-first, minimal chrome, security-terminal aesthetic. Custom CSS overrides remove consumer UX patterns that don't belong in a professional security tool, and the branding communicates clearly that this is a purpose-built system rather than a general-purpose chatbot wrapper.
Design Details
Architecture
OpenWebUI runs as a local web server that connects to VenomX's FastAPI backend via OpenAI-compatible API endpoints. Rather than routing to a cloud provider, the API base URL points to the local vLLM inference server. OpenWebUI handles all conversation state, message history, and streaming while the backend handles retrieval, tool execution, and generation.
Custom OpenWebUI pipelines (Python functions that intercept requests) inject RAG context and format responses before they reach the UI layer.
OpenWebUI's pipeline system allows Python code to intercept requests and responses at the API layer. VenomX uses this to:
Fully local: OpenWebUI, the FastAPI backend, vLLM, and PostgreSQL all run on the same machine. The UI is accessible from the local network with no internet connection required at query time. No credentials, prompts, or tool results leave the host system.
Design Philosophy
VenomX's UI strips away the patterns that work for general-purpose AI products and replaces them with those that work for security professionals.
Streaming token output from vLLM means responses start appearing within ~200ms. No waiting for full generation. Results stream in as the model produces them, keeping workflow momentum.
Every response is formatted for direct use. CVE references are structured, CVSS scores are displayed in context, and tool command output appears in copyable code blocks. Copy-paste to a report, not to another chat window.
The interface stays out of the way. No marketing language, no suggested prompts designed for general audiences, no animated mascots. The chat window, the network map, and the output. Nothing else competes for attention.