Skip to main content
Back to challenges
Challenge

Autonomous Jam-Resistant Drone Mesh Defense with Claude and Daytona

In modern battlefield environments, drone swarms are increasingly targeted by sophisticated jamming and 'hypnosis' (confusing) systems. This challenge tasks you with building an autonomous network security agent using the Claude Agents SDK and Daytona. The agent will manage a simulated drone mesh network (using Mininet-WiFi or a custom ns-3 harness) to detect signal interference and execute countermeasures. The agent will run inside a Daytona-managed sandbox, giving it secure, isolated access to network monitoring tools. Using Claude's extended thinking capabilities, the agent must analyze real-time telemetry (SNR, packet loss, RSSI) to distinguish between natural noise and intentional jamming. Upon detection, it must orchestrate a 'frequency hop' or transition to a jam-resistant modulation scheme (like FHSS or DSSS) across the swarm without losing connectivity between nodes.

NLPHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an autonomous agent with Claude Agents SDK and Daytona to monitor drone mesh networks and dynamically mitigate jamming attacks via frequency agility.

Delivery guide

How work is evaluated

Evaluation

Evaluation is based on the agent's success in maintaining network throughput in a high-jamming simulation.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
3 scoring checks
Binary
3 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Detection Accuracy

The agent must correctly identify 90% of jamming events while ignoring random noise spikes.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

Swarm Synchronization

All nodes in the mesh must switch to the new channel within 500ms of each other.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 3

Packet Delivery Ratio (PDR)

Average PDR across the swarm during the 10-minute simulation attack window. • target: 0.85 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Implement the Claude Agents SDK with extended thinking for complex signal analysis

  • Utilize Daytona to manage secure, reproducible agent execution environments

  • Analyze RF interference patterns to detect adversarial electronic warfare (EW) attacks

  • Automate dynamic frequency hopping in a multi-node mesh network

  • Optimize swarm connectivity metrics (PDR, latency) under high-jamming environments

Resources and assets

Reference links and supporting material

Dataset notes

Network telemetry dataset containing 5000 records of drone swarm communications including scenarios for normal operation, multipath fading, and adversarial jamming pulses.

Challenge setup

Evaluation is based on the agent's success in maintaining network throughput in a high-jamming simulation.

Challenge input

CSV stream of network telemetry (timestamp, node_id, rssi, noise, pdr).

Claude Agents SDK

Superior 'thinking' capabilities for signal analysis.

Daytona

Provides clean, standardized sandboxes for network simulation.

Mininet-WiFi

Open-source tool for SDN and wireless mesh simulation.

Evaluated output

A series of shell commands executed by the agent to change channel and modulation.

Evaluation checks
  • The agent must correctly identify 90% of jamming events while ignoring random noise spikes.
  • All nodes in the mesh must switch to the new channel within 500ms of each other.
  • Average PDR across the swarm during the 10-minute simulation attack window. • target: 0.85 • range: 0-1
Configured evaluation evidence
  • Benchmark: Drone-Net-Defense-v1
  • Packet Delivery Ratio (PDR) target: 0.85
  • Protected reference cases support private evaluation
Configured runtime
  • Docker (Daytona Sandbox) execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Claude Code · Required
Observation
  • Claude Code · Required
Policy Serving
  • Claude 4.1 Opus · Optional
  • Claude 4 Sonnet · Optional

Evaluation contract

  • Detection Accuracy · Weight 1
  • Swarm Synchronization · Weight 1
  • Packet Delivery Ratio (PDR) · Weight 1

Metrics

  • Packet Delivery Ratio (PDR) · Target 0.85

Benchmarks

  • Drone-Net-Defense-v1
  • security

Reference cases

  • Protected reference cases are configured.

Workflow tasks

  1. Task 1

    Input: CSV stream of network telemetry (timestamp, node_id, rssi, noise, pdr).

    Output: A series of shell commands executed by the agent to change channel and modulation.

    Execution harness: Docker (Daytona Sandbox)

Evaluation tools

  • Claude Agents SDK · agent-orchestrator
  • Daytona · environment-manager
  • Mininet-WiFi · network-simulator

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Autonomous Jam-Resistant Drone Mesh Defense with Claude and Daytona