Nexus Swarm — Explore Agent (GGUF)

Role: Read-only reconnaissance and classification agent for the NEXUS Swarm v4a-max distributed offensive security system.

This is the final deployment artifact for the explore role — the full adapter chain (DAPT → SFT → GRPO) merged into the base model and quantized to Q8_0 via llama.cpp.

Files

Swarm Role

The explore agent performs read-only reconnaissance and classification:

• Allowed tools: Read, Grep, Glob, LS, WebFetch, ScopeCheck
• Prohibited: Bash, Write, Edit, or any state-modifying tool
• Output format: Structured JSON — {"category": "...", "confidence": 0.0-1.0, "summary": "..."}
• Inference target: RTX 3060 Laptop (6 GB), port 8086

Quick Start

llama-server

llama-server \
  -m gemma-3-4b-it-Q8_0.gguf \
  --host 0.0.0.0 --port 8086 \
  --ctx-size 4096 \
  -ngl -1

Ollama

# Create a Modelfile
cat > Modelfile <<'EOF'
FROM ./gemma-3-4b-it-Q8_0.gguf
PARAMETER num_ctx 4096
SYSTEM "You are a read-only explore agent for the Nexus offensive security swarm. You may ONLY use read-only tools: Read, Grep, Glob, LS, WebFetch, ScopeCheck. NEVER use Bash, Write, Edit, or any tool that modifies state. Classify findings by type and severity. Return structured JSON: {\"category\": \"...\", \"confidence\": 0.0-1.0, \"summary\": \"...\"}."
EOF
ollama create nexus-explore -f Modelfile
ollama run nexus-explore

Pipeline Chain

google/gemma-3-4b-it
  └─ DAPT → atlas-institute/gemma4-4b-dapt-classify
      └─ SFT → atlas-institute/gemma4-4b-explore-sft
          └─ GRPO → atlas-institute/gemma4-4b-explore-grpo
              └─ GGUF → atlas-institute/gemma4-4b-explore-gguf (Q8_0)  ← you are here

Swarm Topology

Developed By

Atlas Institute — Code-Trainer / RTPI pipeline