Screenshots
Watch trailer on Steam ↗Autonomous Warfare Evolution
SASAKI Naoya · Published by SASAKI Naoya · Official site ↗
1 user reviews — 0% positive (1 reviews)
Last updated 28 Sept 2026
A real-time strategy simulation where you evolve and command swarm-intelligence autonomous weapons using genetic algorithms. Whether you simply watch or deeply intervene in the evolution, both are valid ways to play the game.
£12.49
Lowest price we've ever recorded
View on SteamPrices can lag the live Steam store — always confirm before buying.
Price history
Indie, Simulation, Strategy · 2026 · More £5 - £15 games
About this game
Autonomous Warfare Evolution is a real-time strategy simulation where you evolve and deploy swarm-intelligence autonomous weapons using genetic algorithms.
Swarm Intelligence Evolves Across Generations
The performance and behavioral parameters of swarm-intelligence autonomous weapons—such as vision, attack range, mobility, and formation tactics—are encoded as genomes. Genomes in the pool engage in up to 32 parallel matches, undergoing repeated cycles of selection and crossover to produce the next generation. You can simply sit back and enjoy watching diverse strategies emerge and evolve over time.
Biomes Define Evolutionary Pressure
There are 9 distinct biomes, each with its own environmental parameters. Players can configure evolution-related settings for each biome, including victory conditions and mutation rates, shaping how evolution unfolds.
You Can Control the Swarm
In the combat phase, where evolved genomes are deployed in battle, players can directly influence formations through manual control. However, as AI evolution progresses, human intervention may eventually become ineffective.
Campaign
Open the Mission Console (Tab key / X button on controller) to access two story campaigns: Yamato SAZ Campaign and Tokyo Government Campaign. Each follows a sequence of operations with unlock conditions. Within these campaigns, you can freely experiment with deploying various genomes. Even if defeated, you can incorporate the enemy's genome to evolve further and adapt.
Additionally, the game includes multiple difficulty tiers ("Combat Protocol" generations), where higher tiers feature increasingly evolved AI opponents.
MCP Support
The game supports the Model Context Protocol (MCP), allowing it to be controlled by external AI agents.
You can ask an AI for gameplay advice or let it play through the game directly.
System Requirements
Will my PC run this? →Minimum
- Requires a 64-bit processor and operating system
- OS: Windows 11（64-bit）
- Processor: 1 GHz or faster, 64-bit dual-core CPU or equivalent
- Memory: 4 GB RAM
- Graphics: Integrated GPU
- Storage: 100 MB available space
Recommended
- Requires a 64-bit processor and operating system
Supported Languages
Latest news
Patch NotesOfficial
version 1.1.1
Added support for Agent Plugins 1.0.0 Improved MCP tools Visualized unit attacks on the simulation screen Agent Plugins 1.0.0 is a common packaging standard for AI agent extensions. When the game is launched, it generates the following files: C:\\Users\\ \\AppData\\Local\\AutonomousWarfareEvolution\\agent-plugins\\ autonomous-warfare-evolution/ ├── plugin.json ├── mcp.json └── skills/ └── awe-connection-recovery/ └── SKILL.md When the port is changed in the in-game settings, the port number in mcp.json is updated accordingly. At the moment, this does not provide much benefit to users. However, hosts and plugins that support Agent Plugins have a chicken-and-egg relationship. For now, this game is the egg. As part of the MCP improvements, AI agents can now read reports shown on the map screen. By asking an AI agent, you can get explanations of the game's world in languages other than English. Unit attacks are visualized only on the simulation screen. Attacks are intentionally not visualized during actual battles, as this is considered another form of the fog of war.Read full article on Community Announcements (ssknaoya)AnnouncementsOfficial
Autonomous Warfare Evolution now supports MCP
Added MCP support Added HTTP API support Perhaps this is where Autonomous Warfare truly begins. In 2026, AI agents began shifting from entities that answer to entities that act. The MCP support introduced in v1.1.0 is part of that broader movement. The game can now be controlled by AI. You can ask an AI agent for advice, or let it take control and advance the game directly. Testing was conducted primarily through Codex.Read full article on Community Announcements (ssknaoya)Patch NotesOfficial
version 1.0.10
Added automatic restoration of the previous save if the current save data is corrupted Added a Safe Mode option after an abnormal shutdown Improved multithreading stabilityRead full article on Community Announcements (ssknaoya)Patch NotesOfficial
version 1.0.9
Added the Season Result screen. Thread Count can now be configured to any value between 1 and 32. Thread Count is now automatically configured on first launch. Added a Season Result display duration setting to the System Settings screen. Added a Reset Settings option to the System Settings screen. The word "evolution" is often used as if it were an inherently positive thing. Yet every step forward is built upon countless fallen predecessors. In V1.0.9, I added the Season Result screen. At first glance, it may look like a simple ranking display. However, it also reveals which genomes were reaped and which new genomes emerged to take their place. The flowers that bloom upon the remains of fallen genomes will, in time, be reaped as well. I hope this screen helps convey both the beauty and the cruelty of evolution.Read full article on Community Announcements (ssknaoya)Patch NotesOfficial
version 1.0.8
Added the Genome Analysis screen Updated smoke rendering in combats to reflect real-world wind speed and wind direction trends for each city Added a scan effect to unit rendering during combats Unit performance is represented by genomes, whose actual form is an 18-dimensional vector. Since humans cannot directly visualize data in 18-dimensional space on a 2D plane or in 3D space, dimensionality reduction is applied. The Genome Analysis screen allows selecting among t-SNE, UMAP, and PCA (Principal Component Analysis) as dimensionality reduction algorithms. These methods compress the 18-dimensional vectors into three-dimensional space, enabling clusters of similar genomes to be visualized. Smoke rendering was improved in V1.0.8, but wind does not currently affect battle conditions. If you ever visit these cities, remember the wind from this game. Hamamatsu is windy.Read full article on Community Announcements (ssknaoya)
