thedivergentai/gd-agentic-skills

godot-genre-stealth

Expert blueprint for stealth games (Splinter Cell, Hitman, Dishonored, Thief) covering AI detection systems, vision cones, sound propagation, alert states, light/shadow mechanics, and systemic design.

First seen Feb 10, 2026

Installation

$ npx skills add thedivergentai/gd-agentic-skills --skill godot-genre-stealth

Summary

  • Expert blueprint for stealth games (Splinter Cell, Hitman, Dishonored, Thief) covering AI detection systems, vision cones, sound propagation, alert states, light/shadow mechanics, and systemic design.
  • Use when building stealth-action, tactical infiltration, or immersive sim games requiring enemy awareness systems.
  • Keywords vision cone, detection, alert state, sound propagation, light level, systemic AI, gradual detection.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from thedivergentai/gd-agentic-skills · top by installs.

npx skills add thedivergentai/gd-agentic-skills

Browse all from thedivergentai/gd-agentic-skills

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 678
License LICENSE
Default branch main
Open issues 0
Status Active

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 13,206 B
  • docs SUMMARY.md 452 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 197 installs

SKILL.md

NEVER Do (Expert Anti-Patterns)

Detection & Awareness

  • NEVER use binary "Seen/Not Seen" detection; strictly use a Gradual Detection Meter (0-100%) that builds based on distance, light level, and speed.
  • NEVER use standard RayCast3D nodes for massive amounts of vision checks; strictly use PhysicsDirectSpaceState3D.intersect_ray() to query the PhysicsServer instantly and nodelessly.
  • NEVER allow AI to see through solid geometry; strictly use raycasts between AI eyes and player sample points (Head/Torso/Feet).
  • NEVER use a single sample point for visibility; strictly sample at least 3 points (Head, Torso, Feet) to prevent detection bugs when partially in cover.
  • NEVER use static "Guard Paths"; strictly implement Dynamic Investigating where guards leave their route to check on suspicious sounds/activities.
  • NEVER trigger "Detection" immediately upon line-of-sight; strictly use a Detection Meter with a decay rate to provide a "forgiveness window" for the player to recover.
  • NEVER assume a random navmesh point is safe; strictly verify cover points by Raycasting toward the Threat to ensure geometry successfully breaks the line of sight.
  • NEVER forget to pass the guard's own RID into the raycast exclude array; if omitted, the ray will hit the guard's own body, causing false blocking.
  • NEVER run complex AI detection for off-screen guards; strictly use VisibleOnScreenNotifier3D to pause heavy logic for distant enemies.

Systemic & World Logic

  • NEVER use a simple distance_to() check for hearing; strictly calculate sound travel along the Navigation Path to determine if a wall blocks noise.
  • NEVER make combat as viable as stealth; strictly ensure "going loud" triggers intense reinforcements or high-lethality states to preserve the stealth loop.
  • NEVER hide the "Why" of detection; strictly provide immediate feedback via UI icons (?, !) or audio barks ("What was that?").
  • NEVER ignore the return value of intersectray(); strictly check isempty() first to prevent runtime crashes.
  • NEVER assume a raycast won't hit the guard itself; strictly exclude the guard's RID from Query Parameters.

Optimization & Performance

  • NEVER tightly couple AI to player scripts; strictly use duck-typing (e.g., if body.hasmethod("getdetected")) so guards can spot decoys or dead bodies without brittle dependencies.
  • NEVER maintain hardcoded arrays to trigger base-wide alarms; strictly add guards to a "guards" group and use gettree().callgroup() for dynamic notification.
  • NEVER use standard Strings for AI state; strictly use StringName (&"alert") for O(1) pointer-level comparisons in high-frequency loops.
  • NEVER bake massive NavigationMeshes synchronously; strictly use useasynciterations to prevent main thread stalls during runtime bakes.
  • NEVER rely on Node.find_child() during gameplay; strictly use Groups or exported references for O(1) player tracking.
  • NEVER leave CollisionShapes enabled on incapacitated bodies; strictly disable them or move them to a "corpse" layer to prevent pathing interference.

🛠 Expert Components (scripts/)

MANDATORY reads before implementing the matching system:
1. [stealthpatterns.gd](scripts/stealthpatterns.gd) — space-state LoS, cone DOT, hearing helpers
2. [stealthaicontroller.gd](scripts/stealthaicontroller.gd) — multi-sample rays + RID exclude + path-length hearing
3. [stealthvisioncone.gd](scripts/stealthvisioncone.gd) / [visioncone3d.gd](scripts/visioncone3d.gd) — cone authorship

Original Expert Patterns

  • [stealthpatterns.gd](scripts/stealthpatterns.gd) - Nodeless LoS, cone tests, AI_Audible bus, async investigate.
  • [stealthaicontroller.gd](scripts/stealthaicontroller.gd) - Alert meter AI using PhysicsDirectSpaceState (not RayCast3D nodes).

Modular Components

  • [stealthvisioncone.gd](scripts/stealthvisioncone.gd) - 2D/logic cone helpers.
  • [visioncone3d.gd](scripts/visioncone3d.gd) - 3D cone debug / query helpers.
  • [visibilitymanager.gd](scripts/visibilitymanager.gd) - Player exposure / light gem aggregation.
  • [lightdetector.gd](scripts/lightdetector.gd) - Light probe / overlap exposure.
  • [soundocclusionmanager.gd](scripts/soundocclusionmanager.gd) - Occlusion / bus routing for AI hearing.

Core Loop

  1. Hide / move → 2. Vision & light exposure → 3. Sound propagation → 4. Alert escalation → 5. Investigate / escape

Decision Trees

Perception

Need Action
LoS + exclude self/player RIDs MANDATORY [stealthaicontroller.gd](scripts/stealthaicontroller.gd) + [stealthpatterns.gd](scripts/stealthpatterns.gd)
Light-scaled detection [visibilitymanager.gd](scripts/visibilitymanager.gd) / [lightdetector.gd](scripts/lightdetector.gd)
Hearing through geometry Path-length via NavigationServer (controller) + [soundocclusionmanager.gd](scripts/soundocclusionmanager.gd)

When to load modules

Task Load
Guard AI spine patterns + ai_controller
Author cones vision_cone scripts
Player light gem visibility + light_detector
Loud world props soundocclusionmanager

Do not paste long vision/alert/ability tutorials into the skill body — keep decision trees here and implement from scripts.

Skill Chain

Phase Skills Purpose
1. Physics godot-raycasting-queries Space-state LoS
2. Nav godot-navigation-pathfinding Investigate / path hearing
3. AI godot-state-machine-advanced IDLE→ALERT FSM
4. Audio godot-audio-systems AI_Audible buses
5. Balance godot-monte-carlo-balancer Detection thresholds

Common Pitfalls

Pitfall Solution
RayCast3D per guard Multi-sample intersect_ray + RID exclude
distance_to hearing Navigation path length
Off-screen CPU VisibleOnScreenNotifier suspend

Deep recipes (on demand)

Topic Reference / script
Design pillars [design-principles.md](references/design-principles.md)
Vision / sound / light [ai-detection-system.md](references/ai-detection-system.md) + [stealthaicontroller.gd](scripts/stealthaicontroller.gd)
Alert FSM & UI feedback [alert-states.md](references/alert-states.md) + [visibilitymanager.gd](scripts/visibilitymanager.gd)
Player tools (lean, gadgets) [player-abilities.md](references/player-abilities.md)
Cover & encounter layout [level-design.md](references/level-design.md)
UI communication [ui-communication.md](references/ui-communication.md)

Reference

Progressive disclosure: open Official Documentation links only when researching a specific API; load Related Skills when routing to a peer domain — do not preload the whole lattice.

Official Documentation

  • Ray-casting — Node casts vs PhysicsDirectSpaceState3D.intersect_ray() for many AI LoS checks without RayCast3D spam.
  • Physics introduction — collision layers/masks that filter vision rays, hearing spheres, and corpse layers.
  • PhysicsDirectSpaceState3D — nodeless intersectray / intersect_shape contracts for composite vision and cover verification.
  • PhysicsRayQueryParameters3D — mask, exclude RIDs, and collide flags so guards never self-block LoS queries.
  • Lights and shadows — Omni/Spot energy and shadows that drive light-gem exposure and shadow-layer hiding.
  • Audio buses — dedicated AI-audible buses so loud footsteps/impacts route into hearing systems.
  • Navigation introduction (3D) — regions, agents, and async bake for patrols and investigate paths.
  • Using navigation paths — path length as wall-aware sound travel instead of raw distance_to().
  • NavigationServer3D — mapget_path, random cover points, and avoidance masks for investigating guards.
  • Groups — guards / player / lights groups and call_group for base-wide alarms without hardcoded arrays.
  • VisibleOnScreenNotifier3D — pause heavy detection when off-screen so distant AI do not burn the physics budget.
  • Screen-reading shaders — hintscreentexture vision-cone feedback overlays (not Godot 3 SCREENTEXTURE).

Related Skills

Prerequisites

  • godot-project-foundations — scene tree, groups, and import/project setup before wiring guards, lights, and player sample points.
  • godot-physics-3d — CharacterBody3D, layers/masks, and collision shapes that vision rays and hearing volumes must hit honestly.
  • godot-raycasting-queries — PhysicsServer query parameters, RID excludes, and multi-sample LoS recipes this genre depends on every frame.

Complements

  • godot-navigation-pathfinding — NavigationAgent patrols, investigate targets, and path-length hearing that respects walls.
  • godot-ai-navigation — FOV sensors and perception stacks that feed suspicion/alert state machines.
  • godot-audio-systems — 3D streams and bus layouts for AI-audible noise without coupling to player SFX buses.
  • godot-3d-lighting — light energy, shadows, and probe setups that make light-level detection fair and readable.
  • godot-state-machine-advanced — StringName IDLE/SUSPICIOUS/ALERTED/COMBAT machines instead of ad-hoc string compares.
  • godot-signal-architecture — alertstatechanged and detection-meter signals that keep UI, barks, and AI decoupled.
  • godot-shaders-basics — screen-space cone tint and outline feedback so players always see why they were spotted.
  • godot-camera-systems — lean/peek and frustum-driven VisibleOnScreenNotifier suspend for off-screen guard budgets.
  • godot-monte-carlo-balancer — simulate detection rates, FOV ranges, hearing falloff, and forgiveness windows before shipping difficulty.

Downstream / consumers

  • godot-genre-horror — stalker cones, hiding spots, and suspicion meters that reuse sensory AI patterns.
  • godot-combat-system — lethal escalation and takedown windows once ALERTED/COMBAT breaks the stealth loop.

Master

  • godot-master — library router and mirrored module entry for cross-skill discovery.