Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add IdoCohen560/claude-unity-game-studio --skill unity-npc-behaviorgit clone --depth 1 https://github.com/IdoCohen560/claude-unity-game-studioWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/idocohen560/claude-unity-game-studio/unity-npc-behavior)<a href="https://agentmods.dev/skills/idocohen560/claude-unity-game-studio/unity-npc-behavior"><img src="https://agentmods.dev/badge/skills/idocohen560/claude-unity-game-studio/unity-npc-behavior.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00075 | $0.09886 |
| Opus 5 | $0.00037 | $0.04943 |
| Sonnet 5 | $0.00015 | $0.01977 |
| Haiku 4.5 | $0.00007 | $0.00989 |
Grade A, and why
unity-npc-behavior scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
100% identical to unity-npc-behavior — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 1,272 lines — stays where its author put it; the contents beside it link to each section on GitHub.
NPC Behavior -- Design Translation Patterns
Prerequisite skills:
unity-physics-queries(perception raycasts, NonAlloc query patterns),unity-state-machines(decision FSM/BT architecture),unity-game-architecture(event bus, Service Locator, component composition)
Claude's default NPC is a single MonoBehaviour with an Update() that raycasts, decides, and moves all in one method. This is untestable and unextensible -- changing perception logic requires touching decision logic, adding a new sense means rewriting the entire method, and squad coordination is impossible when each NPC is a self-contained island. These patterns decompose NPC behavior into pluggable layers that designers can configure without programmer intervention.
PATTERN: Perception System Architecture
DESIGN INTENT: NPCs sense the world through sight, hearing, and proximity. Different NPC types have different senses -- a guard has long-range sight but poor hearing; a dog has short sight but exceptional hearing. Senses are mix-and-match components that feed a unified target list.
WRONG (Claude default):
// Raycast inside decision code -- can't swap senses, no batching, untestable
public class EnemyAI : MonoBehaviour
{
void Update()
{
// Perception and decision jammed together
var dir = (player.position - transform.position).normalized;
if (Physics.Raycast(transform.position, dir, out var hit, 20f))
{
if (hit.collider.CompareTag("Player"))
{
// Immediately decide and act -- no separation
Chase(hit.collider.transform);
}
}
}
}
RIGHT:
/// <summary>
/// Pluggable sense interface. Each sense is a separate component on the NPC.
/// Cross-ref: unity-physics-queries for OverlapSphereNonAlloc and raycast patterns.
/// </summary>
public interface IPerceptionSense
{
/// <summary>Evaluate targets within range. Called by PerceptionComponent on staggered schedule.</summary>
void EvaluateTargets(Collider[] scratchBuffer, List<PerceivedTarget> results);
/// <summary>Maximum range of this sense for broad-phase overlap sphere.</summary>
float MaxRange { get; }
}
/// <summary>
/// Central perception hub. Collects results from all IPerceptionSense components.
/// Maintains a deduplicated KnownTargets list for the decision layer.
/// </summary>
public class PerceptionComponent : MonoBehaviour
{
[SerializeField] private float _updateInterval = 0.2f;
[SerializeField] private LayerMask _detectionLayers;
private readonly List<PerceivedTarget> _knownTargets = new();
private IPerceptionSense[] _senses;
private readonly Collider[] _scratchBuffer = new Collider[32];
public IReadOnlyList<PerceivedTarget> KnownTargets => _knownTargets;
}
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 1,272 lines · 75 tokens per session scan A 01a6af737203
unity-npc-behavior is a skill published in the GitHub repository IdoCohen560/claude-unity-game-studio (17 stars, last pushed 2mo ago), licensed MIT. It adds 75 tokens to every session and 9,886 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to unity-npc-behavior, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
swarm
Run a multi-agent audit of a codebase by spawning specialized parallel subagents (security, performance, tests, architecture, dead-code), then synthesize their findings into a single prioritized action plan. Use this whenever the user runs /swarm, asks to "audit the repo," "review this codebase," "find issues across…
memory-loop-setup
Bootstrap the memory loop into any repo or Obsidian vault. Triggers: (1) "set up the memory loop / second brain / warm layer here" or "install this structure in another repo/vault", (2) a new project asking "make the AI remember across sessions and surfaces".
discord-intent-application
Helps write the justification text for Discord Developer Portal privileged intent applications (Message Content, Server Members, Presence) for a specific bot. Takes a plain-text description of the bot's functionality and/or an uploaded/local codebase, screens the bot's features against Discord's known rejection…
plan-qa
Post-implementation QA planning — independent of /review (the developer chooses whether to run them sequentially or in parallel), only when the change has a running surface worth driving. Interviews the developer to turn the specs and the diff into an executable QA specification (specs/qa/QA- - .md) that /execute-qa…
execute-qa
Executes a QA specification (specs/qa/QA- - .md) written by /plan-qa: drives every step with the plan's drivers, verifies each Expected line mechanically ([assert]) or by evidence-backed judgment against the plan's written criterion ([judge]), pauses at named operator handoffs, and writes the results artifact…
generate-tasks
Phase 3 of 5 — slices the ARCH doc into verification-ready task specs (tdd/test-after/ui/checklist), emitted as a separate TASKS- - .md file alongside ARCH for the implement skill. Use only when the user asks to run Phase 3 or generate tasks from an ARCH doc — never trigger automatically.