1
Unified Interface
Local
Deployment
Custom
OpenWebUI Pipelines
0
Cloud Dependencies

OpenWebUI Chat 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.

VenomX · OpenWebUI
You
Run an nmap scan on 192.168.1.0/24 and identify any high-severity CVEs for open services.
VX
✓ Nmap scan complete: 14 hosts up, 47 open ports detected.

Critical findings (CVSS ≥ 9.0):
CVE-2021-44228 · Apache Log4j RCE · 192.168.1.42:8080 · CVSS 10.0
CVE-2022-26134 · Confluence OGNL injection · 192.168.1.55:8090 · CVSS 9.8

Recommended next step: Confirm exploitability with Metasploit module exploit/multi/http/log4shell_header_injection
You
Generate a triage report for the Log4Shell finding.
VX
## Triage Report: CVE-2021-44228
Severity: CRITICAL · CVSS 10.0 (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H)
Affected host: 192.168.1.42 · Apache 2.14.0 on port 8080
MITRE: T1190 Exploit Public-Facing Application · T1059 Command & Scripting
Remediation: Upgrade to Log4j 2.17.1+. Disable JNDI lookups via log4j2.formatMsgNoLookups=true

What the UI Does

Four purpose-built capabilities beyond standard chat, each designed around real security workflows.

Core Feature
Chat Interface

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

  • Natural language to LLM, full conversation history
  • Custom system prompt for security-focused context
  • Markdown rendering for structured report output
  • Code block formatting for exploit snippets and commands
  • Streaming responses all token-by-token displayed without delay
Visualization
Target Management System

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

  • Target records with host, scope, and asset metadata
  • Status and priority indicators for each target
  • Live updates as discovery and assessment progress
  • Clickable targets to drill into findings and notes
Output
Structured Report Renderer

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

  • Executive summary and severity classification
  • CVE / CVSS table with affected hosts
  • MITRE ATT&CK technique mapping
  • Tool output blocks (Nmap, Metasploit, SQLMap)
  • Remediation guidance with version targets
Customization
VenomX Branded Theme

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

  • Dark mode default, persistent light/dark toggle
  • Archivo + Inter + JetBrains Mono type stack
  • VenomX accent blue (#4c88c2) throughout
  • Glassmorphism card surfaces, subtle blur overlays
  • Stripped consumer patterns: no emoji suggestions, no generic onboarding

OpenWebUI Pipeline Integration

How OpenWebUI Connects

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 OpenAI-compatible API vLLM backend Custom pipelines

Pipelines & Filters

OpenWebUI's pipeline system allows Python code to intercept requests and responses at the API layer. VenomX uses this to:

  • Inject RAG-retrieved context into the system prompt before generation
  • Parse tool call results and format them as structured blocks
  • Strip internal reasoning tokens from the final output stream
  • Route scan-related queries to the agent loop vs. direct LLM answers
inlet / outlet filters Context injection Response formatting

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.

Built for Operators, Not Consumers

VenomX's UI strips away the patterns that work for general-purpose AI products and replaces them with those that work for security professionals.

Speed

Low-Latency First

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.

Output

Report-Ready Output

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.

Layout

Minimal Chrome

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.