CORE TopoGen

Generate reproducible CORE network topologies from scenario XML files using a rich Web GUI or a command-line interface.

Table of contents

• Highlights
• Screenshots
• Quick start
  • Prerequisites
  • Install dependencies
  • Launch the Web UI
  • Run the CLI
• Full Preview workflow
• Feature deep dive
  • Planning semantics
  • Router connectivity & aggregation
  • Traffic, segmentation, and services
  • Reports & artifacts
• Architecture overview
• Troubleshooting
• Additional documentation
• Contributing
• Citation

Highlights

• Single-source planning – edit scenarios in the browser or any XML editor and reproduce results with the CLI.
• Deterministic previews – optional RNG seed locks in host expansion, router placement, connectivity, services, segmentation, and vulnerability assignment.
• Live log dock – stream run output, filter by level or text/regex, and toggle auto-follow for long runs.
• Rich topology policies – per-routing-item R2R meshes, R2S aggregation, host grouping bounds, and switch re-homing.
• Artifacts on disk – traffic scripts, segmentation rules, docker-compose definitions, Markdown reports, and JSON summaries are written to predictable locations for inspection.
• Hardware-in-the-Loop friendly – manage HITL attachments directly in the editor; options are constrained to existing routers/switches or provisioning a new router to keep topologies deterministic.

Screenshots

View the WebUI images gallery docs/screenshots.md.

Quick start

Prerequisites

• Python 3.10+ (3.11 recommended)
• CORE 9.2 or newer with core-daemon running
• Docker (optional) for nginx/envoy reverse proxy or vulnerability compose targets

Install dependencies

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Launch the Web UI

python main.py

• Visits http://127.0.0.1:9090 by default (reverse proxy nginx compose targets https://localhost).
• First launch seeds an coreadmin / coreadmin account; change it immediately under Profile → Change Password.
• Configure TLS reverse proxies with make host-web, make host-web-nginx, or make host-web PROXY=envoy. The helper script scripts/dev_gen_certs.sh issues self-signed certificates with SAN support.
• HITL editor note: the “Attach to” dropdown now offers Existing Router, Existing Switch, or New Router. The legacy New Switch choice was removed to prevent hidden switch fan-out; saved scenarios using it are normalized to Existing Router during load.

Run the CLI

python -m core_topo_gen.cli --xml path/to/scenario.xml --seed 42 --verbose

Popular options:

• --scenario NAME pick a specific scenario entry
• --host / --port override the CORE gRPC endpoint (defaults 127.0.0.1:50051)
• --layout-density {compact|normal|spacious} adjust map spacing
• --seg-include-hosts, --seg-allow-docker-ports, --nat-mode, --dnat-prob fine-tune segmentation
• --traffic-pattern, --traffic-rate, --traffic-content override traffic defaults

Full Preview workflow

1. Save XML – The editor auto-saves, but hitting “Save XML” ensures consistent previews.
2. (Optional) Set seed – Enter an integer seed to get deterministic topology output.
3. Generate Full Preview – Shows router/host counts, R2R/R2S policies, segmentation, services, traffic, and vulnerability assignments before any CORE call.
4. Review structured sections – Toggle between structured cards and raw JSON; history stores the last 25 previews in local storage.
5. Run (Seed) – Launches the CLI asynchronously, streams logs into the dock, and writes a Markdown report to ./reports/.

Feature deep dive

Planning semantics

• Host planning honours Base Hosts (density) and Count rows; metadata is written into XML (base_nodes, additive_nodes, combined_nodes, etc.) for round-trip fidelity.
• Router and vulnerability planning capture derived vs explicit counts via explicit_count, derived_count, and total_planned.
• Scenario-level scenario_total_nodes summarises planned hosts, routers, and vulnerability targets.
• Parser helpers expose metadata programmatically: core_topo_gen.parsers.planning_metadata.parse_planning_metadata().
• Hardware-in-the-Loop plans persist per-scenario preferences (enabled state, interface list, attachment choice). Attachments normalize to existing_router, existing_switch, or new_router; legacy new_switch values are coerced to so previews remain deterministic and switch overlays aren’t synthesized implicitly.

Router connectivity & aggregation

• Per-routing-item r2r_mode supports Exact, Uniform, NonUniform, Min, Max, and legacy meshes tied to --router-mesh-style.
• R2S policies (r2s_mode, r2s_edges, optional r2s_hosts_min/max) regroup hosts behind dedicated switches, with “Exact=1” aggregating all hosts per router into a single switch.
• Preview JSON and runtime stats capture router degrees, aggregation counts, and Gini coefficients for quick balance checks.

Traffic, segmentation, and services

• Traffic scripts land in /tmp/traffic (with companion services) and respect overrides for pattern, rate, jitter, and content hints.
• Segmentation scripts land in /tmp/segmentation alongside a segmentation_summary.json; NAT mode, DNAT probability, host inclusion, and docker port allowances are configurable.
• Docker vulnerabilities attach per-node docker-compose files in /tmp/vulns, with metadata embedded into CORE nodes.
• Custom traffic plugins can register via core_topo_gen.plugins.traffic.register() for bespoke sender/receiver code.

Reports & artifacts

• Markdown reports (./reports/scenario_report_<timestamp>.md) enumerate topology stats, planning metadata, segmentation results, and runtime artefacts. Each run also emits a JSON summary alongside the Markdown file (scenario_report_<timestamp>.json) plus per-run connectivity CSVs when router degree data is available.
• Run history is persisted in outputs/run_history.json for the Reports page.
• Safe deletion keeps reports while purging associated outputs under outputs/ when scenarios are removed via the GUI.

Architecture overview

Troubleshooting

• **core-python not found** – set WEBUI_PYbeforemake host-webor rely onpython3; the backend falls back to sys.executable` if needed.
• Empty TLS cert folder (nginx/certs) – run scripts/dev_gen_certs.sh or make dev-certs before composing nginx.
• core-daemon unreachable` – verify daemon status and host/port; GUI run modal will surface connection issues immediately.
• Docker vulnerabilities skipped – ensure images are downloaded/pulled via the Vulnerabilities catalog and Docker is available to the host.
• Log dock won’t auto-scroll – click the “Follow Off/On” toggle to re-enable auto-follow.

Additional documentation

• API.md – REST endpoints exposed by the Web UI backend
• SCENARIO_XML_SCHEMA.md – Schema walkthrough and examples

Contributing

Pull requests and issue reports are welcome! Please run the relevant pytest targets (pytest -q) before submitting changes and keep documentation up to date when behaviour changes.

Cite our work:

@article{rodriguez2025agentcyte,

title={AgentCyTE: Leveraging Agentic AI to Generate Cybersecurity Training & Experimentation Scenarios},

author={Rodriguez, Ana M and Acosta, Jaime and Kotal, Anantaa and Piplai, Aritran},

journal={Annual Computer Security Applications Conference Workshop (ACSACW)},

year={2025}

}