AutoGineer summary
Written by AutoGineer from the skill’s own files, not by the author
Tune Enemy AI provides a structured approach to developing combat artificial intelligence that is predictable for players and testable for developers. By emphasizing explicit state machines, the separation of perception from action, and rigorous testing of decision surfaces, this skill helps creators build enemy behaviors that are fair, readable, and robust. It covers the full lifecycle of AI development, from defining state transitions and movement logic to implementing regression tests for complex combat scenarios.
What it does
- Defines state machine patterns for enemy behaviors like idle, pursue, attack, and retreat.
- Promotes the separation of perception, intent, and motion to ensure consistent AI behavior.
- Provides guidelines for creating fair combat experiences through telegraphing and movement constraints.
- Outlines methods for building deterministic test fixtures to validate AI decision-making.
- Encourages the use of authoritative collision and navigation data over visual assumptions.
When to use it
- When designing new enemy archetypes that require distinct combat personalities.
- When debugging erratic or unfair enemy behavior in an action game.
- When setting up automated regression testing for complex boss encounters or combat states.
- When balancing the difficulty of enemy spacing, attack frequency, and telegraphing.
Advantages
- Improves player experience by making enemy intentions clear and readable.
- Reduces bugs by enforcing explicit state transitions and exit conditions.
- Ensures AI reliability through deterministic testing of decision logic.
- Prevents common AI issues like instant turn-and-hit or perpetual chasing.
Usage examples
- “How should I structure the state machine for a basic melee enemy?”
- The agent provides a state machine model with defined states like idle, pursue, windup, attack, and recover, including specific prerequisites and cooldowns for each transition.
- “How can I test if my boss AI handles player interruptions correctly?”
- The agent guides you to create a deterministic test fixture that simulates an interrupt event during a boss phase and asserts that the AI transitions to the correct stagger state.
Install
Download the skill as a ZIP:
https://mp-staging.autogineer.com/agent-skills/mengto/tune-enemy-ai/downloadUnzip it into the cross-client skills folder of your project:
mkdir -p .agents/skills && unzip tune-enemy-ai.zip -d .agents/skills
Documentation by Meng To
Tune Enemy AI
Make enemy choices legible, bounded, and reproducible.
Model decisions explicitly
Use a small state machine or utility layer with named states such as idle, investigate, pursue, reposition, windup, attack, recover, stagger, retreat, and defeated. State transitions must state their prerequisites, exit conditions, minimum dwell time, and cooldown effects.
Separate perception, intent, and motion
Gather observable inputs: distance, line of sight, target state, occupancy, threat, health, and timers.
Select one intention from constrained legal actions.
Move and animate toward that intent without rewriting the decision mid-action.
Use authoritative collision and navigation results for movement success. Do not derive them from rendered pose or assumed path completion.
Preserve fair combat
Telegraph attacks before their active window. Prevent instant turn-and-hit behavior, perpetual chase, clipped attacks through blockers, and repeated recovery spam. Add spacing and commitment so the player can read and answer each enemy archetype.
Test the decision surface
Create deterministic fixtures for target acquisition, target loss, obstruction, path failure, close-range pressure, multiple enemies, retaliation, interrupt, stagger, boss phase, and reset. Assert transitions and outcomes, not only final positions. Run a real browser encounter after automated tests.
Files
- ATTRIBUTION.mdAdded by AutoGineer854 B
- LICENSELicence file added by AutoGineer1.0 KiB
- SKILL.md1.7 KiB
- agents/openai.yaml195 B
SKILL.md
Tune Enemy AI
Make enemy choices legible, bounded, and reproducible.
Model decisions explicitly
Use a small state machine or utility layer with named states such as idle, investigate, pursue, reposition, windup, attack, recover, stagger, retreat, and defeated. State transitions must state their prerequisites, exit conditions, minimum dwell time, and cooldown effects.
Separate perception, intent, and motion
Gather observable inputs: distance, line of sight, target state, occupancy, threat, health, and timers.
Select one intention from constrained legal actions.
Move and animate toward that intent without rewriting the decision mid-action.
Use authoritative collision and navigation results for movement success. Do not derive them from rendered pose or assumed path completion.
Preserve fair combat
Telegraph attacks before their active window. Prevent instant turn-and-hit behavior, perpetual chase, clipped attacks through blockers, and repeated recovery spam. Add spacing and commitment so the player can read and answer each enemy archetype.
Test the decision surface
Create deterministic fixtures for target acquisition, target loss, obstruction, path failure, close-range pressure, multiple enemies, retaliation, interrupt, stagger, boss phase, and reset. Assert transitions and outcomes, not only final positions. Run a real browser encounter after automated tests.
Requirements and compatibility
Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta
“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.
Security
No issues found by our static scan· No issues found
A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.
Attribution and licence
- Created by
- Meng To
- Publisher
- Meng To (Community)
- Source
- MengTo/Skills · agent-skills/game-development/tune-enemy-ai · commit 5ea1661
- Licence
- MIT License (redistribution allowed)
- LICENSE (added by AutoGineer from LICENSE)
- ATTRIBUTION.md (written by AutoGineer)
- Synced
AutoGineer is not affiliated with or endorsed by Meng To.
Related skills
Web 3d Asset Pipeline
OpenAIOfficial
Prepare and optimize browser-game 3D assets. Use when the user asks for GLB or glTF shipping work, including Blender cleanup and export, collision or LOD setup, compression, texture packaging, and runtime validation.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreLink-onlyGitHub stars of the source repository7.8k
3d Sky Background
Meng To· Community
Guidelines for creating camera-correct, coherent sky backgrounds and environment lighting for 3D scenes.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository7.3k
Design Action Combat
Meng To· Community
A framework for designing and testing readable, state-driven tactical combat systems for web-based games.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository7.3k
Author Game Levels
Meng To· Community
A structured workflow for designing and validating flat-plane Three.js game levels with a focus on readability and deterministic data.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository7.3k
Build Hybrid Game Assets
Meng To· Community
A structured workflow for planning, creating, and integrating hybrid 3D and 2D assets into Three.js web games.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository7.3k
Design Game Encounters
Meng To· Community
A structured workflow for designing, implementing, and balancing action-game encounters within Three.js environments.
Design and creativeClaude CodeClaude appsOpenAI Codex+14 moreMITGitHub stars of the source repository7.3k
Download ZIP
The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download tune-enemy-ai.zip