CCM
/Skills
SkillsMCPMarketplacesDigestToolsAdvertise

This week in Claude

Every Monday: Claude Code, Agent SDK, MCP, and the Anthropic platform moves worth your time.

Skills by Category
Frontend DevelopmentBackend & APIsTesting & QASecurityDevOps & CI/CDGit & Pull RequestsDocumentationCode Review & QualityAI & Agent BuildingSkill Development
MCP Servers by Category
Sales & MarketingWeb & Browser AutomationDatabasesAI & LLM ToolsCloud & InfrastructureCommunication & MessagingDeveloper ToolsDesign & CreativeDocuments & KnowledgeSearch & Web Crawling
Marketplaces by Category
AI Agents & OrchestrationLLM IntegrationDevelopment ToolsFrontend & UIBackend & APIsDatabasesTesting & Code QualityDevOps & CloudSecurity & ComplianceGit & Version Control

Claude Code Marketplaces

Discover Claude Code plugins, extensions, and tools. Automatically updated directory of Anthropic Claude AI marketplaces with development tools, productivity plugins, and integrations.

Resources

  • Browse Skills
  • Browse MCP Servers
  • Browse Marketplaces
  • Skill index
  • MCP index
  • Marketplace index
  • Plugins Reference

Community

  • About
  • Tools
  • Feedback
  • Privacy Policy
  • Advertise

Built for the Claude Code community with Claude Code by mertbuilds.com

Independent project, not affiliated with Anthropic
sentinelcore avatar

Roblox Performance

sentinelcore/roblox-skills
590 installs11 stars
Summary

Covers the full stack of Roblox performance work: StreamingEnabled for large worlds, object pooling for projectiles and effects, caching references outside Heartbeat loops, the task library over legacy wait calls, LOD strategies, and MicroProfiler labeling. The quick reference table and side-by-side bad/good examples make it easy to drop in fixes during a session. It's opinionated in the right ways, like always anchoring static parts and limiting dynamic lights to 10-20 per area. If you're chasing FPS drops or optimizing a laggy server, this gives you the checklist and code patterns without making you dig through DevForum threads.

Install to Claude Code

npx -y skills add sentinelcore/roblox-skills --skill roblox-performance --agent claude-code

Installs into .claude/skills of the current project.

CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
inference shell
inference shell
create and run specialised agents in minutes
build now →
MCP-ready Email SendingMCP-ready Email Sending
MCP-ready Email Sending
Plug Mailtrap into your AI workflow and let it handle the email.
Connect Mailtrap MCP →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Capacitor - Shared memory for your team’s coding agents.
Capacitor - Shared memory for your team’s coding agents.
Make coding agent sessions - Searchable, Shareable, Vendor-neutral & Scored.
Try For Free →
CodeScene MCP ServerCodeScene MCP Server
CodeScene MCP Server
Your agent targets a perfect 10 Code Health score. Deterministic. Every commit.
Try For Free →
Give your AI the whole web as clean markdownGive your AI the whole web as clean markdown
Give your AI the whole web as clean markdown
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
inference shell
inference shell
create and run specialised agents in minutes
build now →
MCP-ready Email SendingMCP-ready Email Sending
MCP-ready Email Sending
Plug Mailtrap into your AI workflow and let it handle the email.
Connect Mailtrap MCP →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Capacitor - Shared memory for your team’s coding agents.
Capacitor - Shared memory for your team’s coding agents.
Make coding agent sessions - Searchable, Shareable, Vendor-neutral & Scored.
Try For Free →
CodeScene MCP ServerCodeScene MCP Server
CodeScene MCP Server
Your agent targets a perfect 10 Code Health score. Deterministic. Every commit.
Try For Free →
Give your AI the whole web as clean markdownGive your AI the whole web as clean markdown
Give your AI the whole web as clean markdown
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Files
SKILL.mdView on GitHub

Roblox Performance Optimization

Quick Reference

TechniqueImpactWhen to Use
StreamingEnabledHighLarge open worlds
Object poolingHighFrequent spawn/destroy
Cache references outside loopsHighHeartbeat/RenderStepped
task.wait() over wait()MediumAll scripts
MeshParts over UnionsMediumMany unique shapes
LOD (hide at distance)MediumComplex models
Anchor static partsMediumReduce physics budget
Limit PointLightsHighAny scene with many lights

StreamingEnabled

Enable for large worlds — engine sends only nearby parts to the client.

-- Studio: Workspace > StreamingEnabled = true
workspace.StreamingEnabled = true
workspace.StreamingMinRadius = 64
workspace.StreamingTargetRadius = 128
  • Parts outside the radius are nil on the client — always guard with if part then.
  • Set Model.LevelOfDetail = Disabled on models that must always be present.
  • Pre-stream an area before a cutscene or teleport:
workspace:RequestStreamAroundAsync(targetPosition, 5) -- 5s timeout

Hot-Path Loop Optimization

RunService.Heartbeat and RenderStepped fire every frame (~60×/sec). Keep them lean.

Bad — searching the hierarchy every frame

RunService.Heartbeat:Connect(function()
    local char = workspace:FindFirstChild(player.Name)
    local humanoid = char and char:FindFirstChild("Humanoid")
    if humanoid then humanoid.WalkSpeed = 16 end
end)

Good — cache references once, do work only when needed

local humanoid = nil

Players.LocalPlayer.CharacterAdded:Connect(function(char)
    humanoid = char:WaitForChild("Humanoid")
end)

RunService.Heartbeat:Connect(function(dt)
    if not humanoid then return end
    humanoid.WalkSpeed = 16  -- cached reference, no search
end)

Rules:

  • Cache game:GetService() and part references outside the loop.
  • Never call FindFirstChild, GetChildren, or GetDescendants inside Heartbeat.
  • Throttle work that doesn't need every frame:
local TICK_INTERVAL = 0.5
local elapsed = 0

RunService.Heartbeat:Connect(function(dt)
    elapsed += dt
    if elapsed < TICK_INTERVAL then return end
    elapsed = 0
    -- expensive work here, runs 2×/sec instead of 60×/sec
end)

task Library vs Legacy Scheduler

Always use task — wait() and spawn() throttle under load and are deprecated.

LegacyModern
wait(n)task.wait(n)
spawn(fn)task.spawn(fn)
delay(n, fn)task.delay(n, fn)
coroutine.wrap(fn)()task.spawn(fn)

Object Pooling

Reuse instances instead of creating and destroying them every frame.

-- ObjectPool ModuleScript
local ObjectPool = {}
ObjectPool.__index = ObjectPool

function ObjectPool.new(template, initialSize)
    local self = setmetatable({ _template = template, _available = {} }, ObjectPool)
    for i = 1, initialSize do
        local obj = template:Clone()
        obj.Parent = nil
        table.insert(self._available, obj)
    end
    return self
end

function ObjectPool:Get(parent)
    local obj = table.remove(self._available) or self._template:Clone()
    obj.Parent = parent
    return obj
end

function ObjectPool:Return(obj)
    obj.Parent = nil
    table.insert(self._available, obj)
end

return ObjectPool
-- Usage
local pool = ObjectPool.new(ReplicatedStorage.Bullet, 20)

local function fireBullet(origin)
    local bullet = pool:Get(workspace)
    bullet.CFrame = CFrame.new(origin)
    task.delay(3, function() pool:Return(bullet) end)
end

Level of Detail (LOD)

Built-in: Set Model.LevelOfDetail = Automatic — engine merges distant parts into an imposter mesh automatically.

Manual distance-based LOD:

-- LocalScript
local INTERVAL = 0.2
local LOD_DISTANCE = 150
local elapsed = 0

RunService.Heartbeat:Connect(function(dt)
    elapsed += dt
    if elapsed < INTERVAL then return end
    elapsed = 0

    local dist = (workspace.CurrentCamera.CFrame.Position - model.PrimaryPart.Position).Magnitude
    local visible = dist < LOD_DISTANCE
    for _, v in model:GetDescendants() do
        if v:IsA("BasePart") then
            v.LocalTransparencyModifier = visible and 0 or 1
        end
    end
end)

Reducing Draw Calls

  • Merge parts that share a material into one MeshPart (export from Blender as .fbx).
  • MeshParts batch better than CSG Unions (Unions re-triangulate at runtime).
  • Reuse materials — 10 parts sharing SmoothPlastic costs far less than 10 unique textures.
  • Use TextureId on a single MeshPart instead of stacking Decals on many parts.

Profiling with MicroProfiler

  1. Press Ctrl+F6 in-game to open MicroProfiler.
  2. Press Ctrl+P to pause and inspect a single frame.
  3. Look for wide bars in heartbeatSignal (Lua), physicsStepped (physics), or render (GPU).
  4. Label your own code:
RunService.Heartbeat:Connect(function()
    debug.profilebegin("MySystem")
    -- your code
    debug.profileend()
end)

Common FPS Killers

CauseFix
Thousands of individual partsMerge into MeshParts
Unanchored static geometryAnchored = true on anything that never moves
Many PointLight / SpotLight instancesLimit to ~10–20 dynamic lights per area
High-rate ParticleEmittersLower Rate and Lifetime; disable when off-screen
wait() under heavy loadReplace with task.wait()
FindFirstChild chains inside HeartbeatCache on load
StreamingEnabled off on large mapsEnable it
Model.LevelOfDetail = Disabled everywhereUse Automatic where safe

Common Mistakes

MistakeFix
workspace:FindFirstChild every frameCache reference on character/model load
Destroying and re-creating bullets/effectsUse an object pool
wait() in tight loopstask.wait()
All parts with unique materialsStandardize to a small set of shared materials
ParticleEmitters enabled off-screenDisable Enabled when particle source is not visible
Physics on decorative partsAnchored = true
Featured
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
inference shell
inference shell
create and run specialised agents in minutes
build now →
MCP-ready Email SendingMCP-ready Email Sending
MCP-ready Email Sending
Plug Mailtrap into your AI workflow and let it handle the email.
Connect Mailtrap MCP →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Capacitor - Shared memory for your team’s coding agents.
Capacitor - Shared memory for your team’s coding agents.
Make coding agent sessions - Searchable, Shareable, Vendor-neutral & Scored.
Try For Free →
CodeScene MCP ServerCodeScene MCP Server
CodeScene MCP Server
Your agent targets a perfect 10 Code Health score. Deterministic. Every commit.
Try For Free →
Give your AI the whole web as clean markdownGive your AI the whole web as clean markdown
Give your AI the whole web as clean markdown
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Categories
Backend & APIsDebuggingGame Development
First SeenJun 3, 2026
View on GitHub

More from sentinelcore/roblox-skills

All 6 skills →
  • Roblox Security494
  • Roblox Datastores429
  • Roblox Remote Events394
  • Roblox Gui815
  • Roblox Animations611

Recommended

More Backend & APIs →
aliramw avatar
dingtalk-ai-table

aliramw/dingtalk-ai-table

钉钉 AI 表格(多维表)操作技能。使用 mcporter CLI 连接钉钉官方新版 AI 表格 MCP server,基于 baseId / tableId / fieldId / recordId 体系执行 Base、Table、Field、Record 的查询与增删改。适用于创建 AI 表格、搜索表格、读取表结构、批量增删改记录、批量建字段、更新字段配置、按模板建表等场景。需要配置 DINGTALK_MCP_URL 或直接使用 Streamable HTTP URL。
589
108
freestylefly avatar
seedance-video

freestylefly/canghe-skills

使用字节跳动 Seedance 模型生成视频。支持文生视频和图生视频功能,通过 volcengine-ark SDK 调用 API。当用户需要生成视频、创建视频内容或基于文字/图片制作视频时激活此技能。
586
429
vm0-ai avatar
strava

vm0-ai/vm0-skills

Strava API for fitness activities. Use when user mentions "Strava", "running", "cycling", "activity", or asks about fitness tracking.
585
76
mlflow avatar
retrieving-mlflow-traces

mlflow/skills

Retrieves MLflow traces using CLI or Python API. Use when the user asks to get a trace by ID, find traces, filter traces by status/tags/metadata/execution time, query traces, or debug failed traces. Triggers on "get trace", "search traces", "find failed traces", "filter traces by", "traces slower than", "query MLflow traces".
584
69
orchestra-research avatar
fine-tuning-openvla-oft

orchestra-research/ai-research-skills

Fine-tunes and evaluates OpenVLA-OFT and OpenVLA-OFT+ policies for robot action generation with continuous action heads, LoRA adaptation, and FiLM conditioning on LIBERO simulation and ALOHA real-world setups. Use when reproducing OpenVLA-OFT paper results, training custom VLA action heads (L1 or diffusion), deploying server-client inference for ALOHA, or debugging normalization, LoRA merge, and cross-GPU issues.
584
11.5k
mindrally avatar
netlify-development

mindrally/skills

Netlify development best practices for serverless functions, edge functions, Blobs storage, build configuration, and deployment workflows.
582
223