Vigil v1.1 — The Security Agency That Never Sleeps
Massive update just dropped. Here's everything that shipped today.
——— VIGIL BRAIN ———
We gave Vigil a brain. Not a wrapper around an LLM — an actual embedded security knowledge base with 356 pre-built entries that answers instantly without calling any AI provider.
What's inside:
• 85 MITRE ATT&CK techniques across all 14 tactics — with real CVEs, detection methods, and
mitigations
• 10 OWASP Top 10 Web risks with CWE mappings and exploitation details
• 10 OWASP LLM Top 10 MITRE ATLAS for AI/ML security
• 48 NIST controls (CSF 2.0 800-53 Rev 5)
• 40 CompTIA Security domains covering the full SY0-701 exam scope
• 89 port-to-service mappings with known vulnerabilities (port 21 through 50000)
• 30 cross-framework compliance mappings (PCI DSS 4.0, HIPAA, SOC 2, ISO 27001, CIS v8)
• 17 CVE vulnerability patterns (SQLi, XSS, SSRF, deserialization, buffer overflow...)
• 27 remediation templates with language-specific fixes for Node.js, Python, Java, Go, PHP
Ask "What is T1059?" — instant answer. "Port 445?" — instant. "CWE-89?" — instant. No LLM call needed. Under 1ms.
When you DO ask something that needs reasoning, the Brain enriches the prompt with your profile context, section awareness, recalled memories, relevant KB entries, and suggested actions — then sends it to your AI provider with full context.
The Brain also remembers. It extracts IPs, domains, CVEs, and preferences from your conversations and recalls them in future sessions. It builds a security profile of your infrastructure — cloud providers, compliance frameworks, threat model, crown jewels — and tailors every response to your environment.
——— FLOATING BRAIN CHATBOX ———
New draggable floating panel — the orange shield button in the bottom right. Grab it and drag it any where on screen. Click it to open a quick-question panel that's section-aware. Navigate to Port Scanner and the panel suggests "What ports should I check?" Navigate to Compliance and it suggests
"How do I prepare for a SOC 2 audit?"
Keyboard shortcut: Ctrl Shift B.
The Brain is wired into the existing AI Terminal and AI Chat — not a separate page. Every AI call through Vigil now goes through the Brain first.
——— FLOWS ———
DAG-based workflow automation is live.
Build security pipelines by chaining nodes together:
• Node types: start, end, LLM, agent, tool, condition, loop, HTTP, delay, human input, notify
• 4 built-in templates: Recon Pipeline, Compliance Check, Incident Response, Vulnerability Triage
• Visual flow editor with drag-drop node placement and edge drawing
• Conditional branching with state evaluation
• Real-time execution progress via
Socket.IO
• PostgreSQL persistence with versioning
Think "scan target → check vulns → if critical → alert Slack → create ticket" — fully automated.
——— AGENT EDITOR 28 SECURITY AGENTS ———
Full agent system with create, edit, and custom system prompts.
4 categories, 28 built-in agents:
Scanners (7): Port Scanner, Subdomain Enumerator, HTTP Header Auditor, XSS Scanner, SQL Injection Detector, TLS Analyzer, Prompt Injection Tester
Analyzers (6): AWS Security Auditor, IAM Policy Analyzer, PCI DSS Checker, HIPAA Compliance Checker, Data Classifier, AI Threat Analyst
Defenders (4): Incident Playbook Generator, Firewall Rule Auditor, Malware Behavior Analyzer, Patch Reviewer
Hunters (9): Log Threat Hunter, Network Anomaly Detector, Memory Forensics, Disk Forensics, Adversarial Analyst, Exploit Validator, Attack Path Mapper, Red Team Planner, Autonomous Pentester
The Autonomous Pentester runs a P-E-R (Planner-Executor-Reflector) cycle with dual causal graph reasoning. The Adversarial Analyst uses the MUST-GATE framework. Not toy agents — these have serious methodology behind them.
Every agent supports per-agent AI provider selection. Run your hunters on local Ollama, your analyzers on Claude API. Your choice.
——— SMART PROVIDER ROUTING ———
Multi-provider AI with intelligent routing:
• Providers: Claude API, Claude CLI, Codex CLI, Ollama (local)
• Strategies: balanced, premium, speed, economy
• Route-based assignment: scans → Ollama, analysis → Claude API, hunts → Ollama
• Per-agent and per-flow provider pinning
• Automatic fallback chains
——— OLLAMA KALI AIR-GAPPED ———
Full Docker stack: Vigil PostgreSQL Ollama.
Default model: qwen3:8b. Runs entirely local. Zero data leaves your machine. Works on Kali Linux.
Works air-gapped. Works on a laptop in a SCIF.
Docker Compose up and you have a complete security operations platform with AI in under 5 minutes.
The security bridge container runs nmap, nuclei, nikto, sqlmap, gobuster, subfinder, httpx, whois, dnsrecon, hydra, and dirb — all sandboxed with input validation and 50MB buffer limits.
——— WHAT ELSE SHIPPED ———
• Glass card visibility fix — surfaces bumped from 65% to 92% opacity, borders and text brightened
across the board
• 2FA hardened — TOTP with challenge-token flow, inline forms replacing window.prompt
• Auth hardening — bootstrap passwords, RBAC enforcement
• Claude CLI in Docker — entrypoint fixes, credential mounting from host
• Express 5.2.1 upgrade
• 40 sidebar sections all wired to Brain context
——— WHAT'S NEXT ———
Vigil is open source. Self-hosted. Your data stays on your infrastructure.
28 agents pre-built. 356 knowledge base entries. 89 port mappings. 122 MITRE techniques. 160 CWE mappings. 30 compliance cross-references. Instant answers.
The security agency that never sleeps.
github.com/vigil-agency/vigi…
#cybersecurity #infosec #opensecurity #MITREATTACK #OWASP #pentesting #threatintel #selfhosted #ollama #kalilinux #secops #redteam #blueteam #purpleteam