Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Build Three.js Scroll Worlds · mengto | DecimalAI

Frontendby mengto

Build Three.js Scroll Worlds

Build rich, scroll-controlled real-time Three.js experiences as one persistent 3D world whose camera, lighting, atmosphere, materials, objects, DOM story, and interactions evolve across authored chapters.

Never made a case worseMaintained

SkillScore

SkillScore87/100

Test results 87

Safety

3/3 safety checks passed

Code scan
Intent
Content

What it changes

83%87%of test cases passed +4 pts

Without skill19 of 23

With skill20 of 23

1 fixed19 passed either way3 still failing0 broke

Measured on gemini-3.6-flash8/29/2026v1

Track record

Agent runs: 0; All time

See one task, with and without the skill

One test taskWe are implementing a custom scroll controller for a multi-stop 3D WebGL web experience. Should we accumulate delta values from wheel events to move the camera, or should we query document scroll metrics? How should scroll progress be passed to render loops versus navigation logic?

Without the skillMissed

With the skillPassed

Same model, same task. Answers are capped and redacted.

Use this skill

Your agent runs on

Adds every file the skill has, pinned so you can update it. No account needed.

pip install "decimalai>=0.15"
decimalai skills add mengto-build-threejs-scroll-worlds
claude --plugin-dir "$(git rev-parse --show-toplevel 2>/dev/null || pwd)/.agents"

The same install from Python, for a notebook or a setup script.

Install

pip install "decimalai>=0.15"

Add the skill

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

print("Added. Start Claude Code with the command below.")

Start Claude Code with it

claude --plugin-dir "$(git rev-parse --show-toplevel 2>/dev/null || pwd)/.agents"

Type these in a Claude Code session. No account or API key. It writes the skill and any files it bundles, and asks first when its safety rating isn't Passed.

Add the DecimalAI plugin, once

/plugin marketplace add decimal-labs/decimalai-claude-plugins
/plugin install decimalai@decimalai

Install this skill

/decimalai:install mengto-build-threejs-scroll-worlds

The plugin needs python3 and curl on your machine.

No Python or account. Claude Code picks it up in its next session.

Save it into your project

curl -fsSL --create-dirs -o .claude/skills/mengto-build-threejs-scroll-worlds/SKILL.md https://app.decimal.ai/s/mengto-build-threejs-scroll-worlds/SKILL.md

Copy SKILL.md

Save it in your project, as

.claude/skills/mengto-build-threejs-scroll-worlds/SKILL.md

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Claude Code docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[claude-agent-sdk]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import asyncio

from claude_agent_sdk import ClaudeAgentOptions
from decimalai.claude_agent_sdk import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

async def main():
    question = "Hello!"
    async with support.run(question) as run:  # one user message, one run
        async for message in run.query(options=ClaudeAgentOptions(max_turns=6)):
            print(message)

# In a notebook or async code: await main()
asyncio.run(main())

Prefer its own loader? ClaudeAgentOptions(plugins=[skills_plugin()], skills="all") reads the same folder. Use one or the other: a skill delivered both ways reaches the prompt twice.

The same install from Python, for a notebook or a setup script. agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[claude-agent-sdk]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import asyncio

from claude_agent_sdk import ClaudeAgentOptions
from decimalai.claude_agent_sdk import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

async def main():
    question = "Hello!"
    async with support.run(question) as run:  # one user message, one run
        async for message in run.query(options=ClaudeAgentOptions(max_turns=6)):
            print(message)

# In a notebook or async code: await main()
asyncio.run(main())

Prefer its own loader? ClaudeAgentOptions(plugins=[skills_plugin()], skills="all") reads the same folder. Use one or the other: a skill delivered both ways reaches the prompt twice.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

import asyncio
from pathlib import Path
from claude_agent_sdk import ClaudeAgentOptions, query

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
options = ClaudeAgentOptions(system_prompt=f"You are a helpful agent.\n\n{skill}", max_turns=6)

async def main():
    async for message in query(prompt="Hello!", options=options):
        print(message)

# In a notebook or async code: await main()
asyncio.run(main())

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Claude Agent SDK docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[openai-agents]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from agents import Agent, Runner
from decimalai.openai_agents import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = Agent(
    name=support.name,
    instructions=support.instructions,  # your prompt, then the skill list
    tools=[*support.tools],             # your tools, then load_skill
)
# In a notebook or async code: result = await Runner.run(agent, "Hello!")
result = Runner.run_sync(agent, "Hello!")

Using SandboxAgent? capabilities=[Shell(), decimalai.openai_agents.sandbox_skills()] points its own loader at the same folder.

Install

pip install "decimalai[openai-agents]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from agents import Agent, Runner
from decimalai.openai_agents import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = Agent(
    name=support.name,
    instructions=support.instructions,  # your prompt, then the skill list
    tools=[*support.tools],             # your tools, then load_skill
)
# In a notebook or async code: result = await Runner.run(agent, "Hello!")
result = Runner.run_sync(agent, "Hello!")

Using SandboxAgent? capabilities=[Shell(), decimalai.openai_agents.sandbox_skills()] points its own loader at the same folder.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from agents import Agent, Runner

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = Agent(name="support", instructions=f"You are a helpful agent.\n\n{skill}")
# In a notebook or async code: result = await Runner.run(agent, "Hello!")
result = Runner.run_sync(agent, "Hello!")

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

OpenAI Agents docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[openai]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import json

from openai import OpenAI
from decimalai.openai import agent_config

client = OpenAI()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")


def lookup_order(order_id: str) -> str:
    return f"Order {order_id} shipped yesterday."


# Your tools, as Chat Completions takes them, and what answers each.
TOOLS = [{"type": "function", "function": {
    "name": "lookup_order",
    "description": "Look up the status of an order.",
    "parameters": {"type": "object", "properties": {"order_id": {"type": "string"}},
                   "required": ["order_id"]},
}}]
FUNCTIONS = {"lookup_order": lookup_order}

question = "Where is order A1001?"
with support.run(question) as run:  # one user message, one run
    # run.system is your prompt, then the skill list.
    messages = [{"role": "system", "content": run.system}, run.message]
    while True:
        reply = client.chat.completions.create(
            model="gpt-5.4-mini",
            messages=messages,
            tools=[*TOOLS, *run.tools],  # your tools, then load_skill
        ).choices[0].message
        messages.append(reply)
        if not reply.tool_calls:
            break
        for call in reply.tool_calls:
            # load_skill is answered by the run; your own tools by FUNCTIONS.
            function = FUNCTIONS.get(call.function.name)
            messages.append(run.tool_result(call) or {
                "role": "tool", "tool_call_id": call.id,
                "content": function(**json.loads(call.function.arguments)) if function
                else f"There is no tool named {call.function.name!r}.",
            })
print(reply.content)

# The Responses API takes run.response_tools instead of run.tools.

Install

pip install "decimalai[openai]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import json

from openai import OpenAI
from decimalai.openai import agent_config

client = OpenAI()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")


def lookup_order(order_id: str) -> str:
    return f"Order {order_id} shipped yesterday."


# Your tools, as Chat Completions takes them, and what answers each.
TOOLS = [{"type": "function", "function": {
    "name": "lookup_order",
    "description": "Look up the status of an order.",
    "parameters": {"type": "object", "properties": {"order_id": {"type": "string"}},
                   "required": ["order_id"]},
}}]
FUNCTIONS = {"lookup_order": lookup_order}

question = "Where is order A1001?"
with support.run(question) as run:  # one user message, one run
    # run.system is your prompt, then the skill list.
    messages = [{"role": "system", "content": run.system}, run.message]
    while True:
        reply = client.chat.completions.create(
            model="gpt-5.4-mini",
            messages=messages,
            tools=[*TOOLS, *run.tools],  # your tools, then load_skill
        ).choices[0].message
        messages.append(reply)
        if not reply.tool_calls:
            break
        for call in reply.tool_calls:
            # load_skill is answered by the run; your own tools by FUNCTIONS.
            function = FUNCTIONS.get(call.function.name)
            messages.append(run.tool_result(call) or {
                "role": "tool", "tool_call_id": call.id,
                "content": function(**json.loads(call.function.arguments)) if function
                else f"There is no tool named {call.function.name!r}.",
            })
print(reply.content)

# The Responses API takes run.response_tools instead of run.tools.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from openai import OpenAI

client = OpenAI()
skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
reply = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[
        {"role": "system", "content": f"You are a helpful agent.\n\n{skill}"},
        {"role": "user", "content": "Hello!"},
    ],
).choices[0].message

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

OpenAI SDK docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[anthropic]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import anthropic
from decimalai.anthropic import agent_config

client = anthropic.Anthropic()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

question = "Hello!"
with support.run(question) as run:  # one user message, one run
    answer = client.beta.messages.tool_runner(
        model="claude-haiku-5-5",
        max_tokens=4096,
        system=run.system,     # your prompt, then the skill list
        tools=[*run.tools],    # your tools, then load_skill
        messages=[{"role": "user", "content": question}],
    ).until_done()

Install

pip install "decimalai[anthropic]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import anthropic
from decimalai.anthropic import agent_config

client = anthropic.Anthropic()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

question = "Hello!"
with support.run(question) as run:  # one user message, one run
    answer = client.beta.messages.tool_runner(
        model="claude-haiku-5-5",
        max_tokens=4096,
        system=run.system,     # your prompt, then the skill list
        tools=[*run.tools],    # your tools, then load_skill
        messages=[{"role": "user", "content": question}],
    ).until_done()

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
import anthropic

client = anthropic.Anthropic()
skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
reply = client.messages.create(
    model="claude-haiku-5-5",
    max_tokens=4096,
    system=f"You are a helpful agent.\n\n{skill}",
    messages=[{"role": "user", "content": "Hello!"}],
)

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Anthropic docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[google-genai]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from google import genai
from google.genai import types
from decimalai.google_genai import agent_config

client = genai.Client()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")


def lookup_order(order_id: str) -> str:
    return f"Order {order_id} shipped yesterday."


# Your functions, as declarations for the model, and what answers each.
TOOLS = [{"function_declarations": [{
    "name": "lookup_order",
    "description": "Look up the status of an order.",
    "parameters_json_schema": {"type": "object", "properties": {"order_id": {"type": "string"}},
                               "required": ["order_id"]},
}]}]
FUNCTIONS = {"lookup_order": lookup_order}

question = "Where is order A1001?"
with support.run(question) as run:  # one user message, one run
    config = types.GenerateContentConfig(
        system_instruction=run.system,     # your prompt, then the skill list
        tools=[*TOOLS, *run.tool_params],  # your functions, then load_skill
    )
    contents = [{"role": "user", "parts": [{"text": run.message}]}]
    while True:
        response = client.models.generate_content(
            model="gemini-3.5-flash", contents=contents, config=config,
        )
        if not response.function_calls:
            break
        contents.append(response.candidates[0].content)
        parts = []
        for call in response.function_calls:
            # load_skill is answered by the run; your own functions by FUNCTIONS.
            # A loop of your own rather than automatic function calling, which
            # raises KeyError when the model asks for a function you did not pass.
            function = FUNCTIONS.get(call.name)
            parts.append(run.tool_result(call) or {"function_response": {
                "name": call.name, "id": call.id,
                "response": {"result": function(**(call.args or {})) if function
                             else f"There is no function named {call.name!r}."},
            }})
        contents.append({"role": "user", "parts": parts})
print(response.text)

Install

pip install "decimalai[google-genai]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from google import genai
from google.genai import types
from decimalai.google_genai import agent_config

client = genai.Client()
# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")


def lookup_order(order_id: str) -> str:
    return f"Order {order_id} shipped yesterday."


# Your functions, as declarations for the model, and what answers each.
TOOLS = [{"function_declarations": [{
    "name": "lookup_order",
    "description": "Look up the status of an order.",
    "parameters_json_schema": {"type": "object", "properties": {"order_id": {"type": "string"}},
                               "required": ["order_id"]},
}]}]
FUNCTIONS = {"lookup_order": lookup_order}

question = "Where is order A1001?"
with support.run(question) as run:  # one user message, one run
    config = types.GenerateContentConfig(
        system_instruction=run.system,     # your prompt, then the skill list
        tools=[*TOOLS, *run.tool_params],  # your functions, then load_skill
    )
    contents = [{"role": "user", "parts": [{"text": run.message}]}]
    while True:
        response = client.models.generate_content(
            model="gemini-3.5-flash", contents=contents, config=config,
        )
        if not response.function_calls:
            break
        contents.append(response.candidates[0].content)
        parts = []
        for call in response.function_calls:
            # load_skill is answered by the run; your own functions by FUNCTIONS.
            # A loop of your own rather than automatic function calling, which
            # raises KeyError when the model asks for a function you did not pass.
            function = FUNCTIONS.get(call.name)
            parts.append(run.tool_result(call) or {"function_response": {
                "name": call.name, "id": call.id,
                "response": {"result": function(**(call.args or {})) if function
                             else f"There is no function named {call.name!r}."},
            }})
        contents.append({"role": "user", "parts": parts})
print(response.text)

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from google import genai
from google.genai import types

client = genai.Client()
skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
response = client.models.generate_content(
    model="gemini-3.5-flash",
    contents="Hello!",
    config=types.GenerateContentConfig(system_instruction=f"You are a helpful agent.\n\n{skill}"),
)

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Gemini API docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[langchain]" langchain-openai
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from langchain.agents import create_agent
from decimalai.langchain import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = create_agent(
    "openai:gpt-5.4-mini",
    tools=[],                         # your tools; the middleware adds load_skill
    middleware=[support.middleware],  # your prompt, the skill list and load_skill
)
result = agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

Install

pip install "decimalai[langchain]" langchain-openai

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from langchain.agents import create_agent
from decimalai.langchain import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = create_agent(
    "openai:gpt-5.4-mini",
    tools=[],                         # your tools; the middleware adds load_skill
    middleware=[support.middleware],  # your prompt, the skill list and load_skill
)
result = agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from langchain.agents import create_agent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = create_agent("openai:gpt-5.4-mini", tools=[], system_prompt=f"You are a helpful agent.\n\n{skill}")
result = agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

LangChain docs

CrewAI's own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" "crewai>=1.15.27"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import decimalai.crewai
from crewai import Agent, Crew, Task

agent = Agent(
    role="Support", goal="Help the customer", backstory="A support agent.",
    llm=llm,                               # your own CrewAI LLM
    skills=decimalai.crewai.skills(),      # the installed folder
    tools=decimalai.crewai.skill_tools(),  # reads a skill's files, runs its scripts
)
task = Task(description="Hello!", expected_output="an answer", agent=agent)
Crew(agents=[agent], tasks=[task]).kickoff()

The same install from Python, for a notebook or a setup script. CrewAI's own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" "crewai>=1.15.27"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import decimalai.crewai
from crewai import Agent, Crew, Task

agent = Agent(
    role="Support", goal="Help the customer", backstory="A support agent.",
    llm=llm,                               # your own CrewAI LLM
    skills=decimalai.crewai.skills(),      # the installed folder
    tools=decimalai.crewai.skill_tools(),  # reads a skill's files, runs its scripts
)
task = Task(description="Hello!", expected_output="an answer", agent=agent)
Crew(agents=[agent], tasks=[task]).kickoff()

No DecimalAI package. CrewAI's own loader reads the folder, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from crewai import Agent, Crew, Task

agent = Agent(
    role="Support", goal="Help the customer", backstory="A support agent.",
    llm=llm,                         # your own CrewAI LLM
    skills=[Path(".agents/skills")],
)
task = Task(description="Hello!", expected_output="an answer", agent=agent)
Crew(agents=[agent], tasks=[task]).kickoff()

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

CrewAI docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[adk]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from decimalai.adk import DecimalAIPlugin, agent_config
from google.adk.agents import LlmAgent
from google.adk.apps import App

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = LlmAgent(
    name="support",          # ADK needs a Python identifier
    model="gemini-3.5-flash",
    instruction=support.instruction,  # your prompt, then the skill list
    tools=[*support.tools],           # your tools, then load_skill
)
# DecimalAIPlugin makes each invocation of runners built from this App one trace.
app = App(name="support", root_agent=agent, plugins=[DecimalAIPlugin()])

Prefer ADK's own loader? SkillToolset(skills=decimalai.adk.skills(), ...) reads the same folder. Use one or the other: a skill delivered both ways reaches the prompt twice.

Install

pip install "decimalai[adk]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from decimalai.adk import DecimalAIPlugin, agent_config
from google.adk.agents import LlmAgent
from google.adk.apps import App

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = LlmAgent(
    name="support",          # ADK needs a Python identifier
    model="gemini-3.5-flash",
    instruction=support.instruction,  # your prompt, then the skill list
    tools=[*support.tools],           # your tools, then load_skill
)
# DecimalAIPlugin makes each invocation of runners built from this App one trace.
app = App(name="support", root_agent=agent, plugins=[DecimalAIPlugin()])

Prefer ADK's own loader? SkillToolset(skills=decimalai.adk.skills(), ...) reads the same folder. Use one or the other: a skill delivered both ways reaches the prompt twice.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from google.adk.agents import LlmAgent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = LlmAgent(
    name="support",
    model="gemini-3.5-flash",
    instruction=lambda context: f"You are a helpful agent.\n\n{skill}",
)

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Google ADK docs

agent_config() hands your agent the skill. With no DecimalAI key it reads your project's files and sends nothing. Needs SDK 0.27 or newer.

Install

pip install "decimalai[pydantic-ai]"
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from pydantic_ai import Agent
from decimalai.pydantic_ai import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = Agent(
    "openai:gpt-4o-mini",
    name=support.name,
    tools=[],                 # your tools; the capability adds load_skill
    capabilities=[support],   # your prompt, the skill list and load_skill
)
# In a notebook or async code: result = await agent.run("Hello!")
result = agent.run_sync("Hello!")

On the Pydantic AI harness, decimalai.pydantic_ai_harness.skills(project) uses its own loader instead.

Install

pip install "decimalai[pydantic-ai]"

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from pydantic_ai import Agent
from decimalai.pydantic_ai import agent_config

# With no DECIMAL_API_KEY this reads the skills your project lists, from disk.
support = agent_config("support", fallback="You are a helpful agent.")

agent = Agent(
    "openai:gpt-4o-mini",
    name=support.name,
    tools=[],                 # your tools; the capability adds load_skill
    capabilities=[support],   # your prompt, the skill list and load_skill
)
# In a notebook or async code: result = await agent.run("Hello!")
result = agent.run_sync("Hello!")

On the Pydantic AI harness, decimalai.pydantic_ai_harness.skills(project) uses its own loader instead.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from pydantic_ai import Agent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = Agent("openai:gpt-4o-mini", instructions=f"You are a helpful agent.\n\n{skill}")
# In a notebook or async code: result = await agent.run("Hello!")
result = agent.run_sync("Hello!")

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Pydantic AI docs

Deep Agents' own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" deepagents
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import decimalai.deepagents
from deepagents import create_deep_agent

agent = create_deep_agent(model=model, **decimalai.deepagents.skills())
agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

The default backend reads files but has no shell. To run a skill's scripts, pass skills(backend=LocalShellBackend(...)).

The same install from Python, for a notebook or a setup script. Deep Agents' own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" deepagents

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import decimalai.deepagents
from deepagents import create_deep_agent

agent = create_deep_agent(model=model, **decimalai.deepagents.skills())
agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

The default backend reads files but has no shell. To run a skill's scripts, pass skills(backend=LocalShellBackend(...)).

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from deepagents import create_deep_agent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = create_deep_agent(model=model, system_prompt=f"You are a helpful agent.\n\n{skill}")
agent.invoke({"messages": [{"role": "user", "content": "Hello!"}]})

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Deep Agents docs

Its own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" agent-framework
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

import asyncio

from agent_framework import Agent
from decimalai.agent_framework import skills

agent = Agent(client=client, instructions="You are a helpful agent.", context_providers=[skills()])
asyncio.run(agent.run("Hello!"))

It caches skills by default, so restart the process after decimalai skills update.

The same install from Python, for a notebook or a setup script. Its own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" agent-framework

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

import asyncio

from agent_framework import Agent
from decimalai.agent_framework import skills

agent = Agent(client=client, instructions="You are a helpful agent.", context_providers=[skills()])
asyncio.run(agent.run("Hello!"))

It caches skills by default, so restart the process after decimalai skills update.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

import asyncio
from pathlib import Path
from agent_framework import Agent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = Agent(client=client, instructions=f"You are a helpful agent.\n\n{skill}")
asyncio.run(agent.run("Hello!"))

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Microsoft Agent Framework docs

Strands' own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" strands-agents
decimalai skills add mengto-build-threejs-scroll-worlds

Add it to your agent

from decimalai.strands import skills
from strands import Agent
from strands.vended_tools import file_editor, shell

agent = Agent(
    model=model,                # always pass one: the default is Amazon Bedrock
    system_prompt="You are a helpful agent.",
    tools=[file_editor, shell],
    plugins=[skills()],
)
agent("Hello!")

Strands loads an agent's skills once, so build the agent again after decimalai skills update.

The same install from Python, for a notebook or a setup script. Strands' own skill loader reads the installed folder. No DecimalAI account needed.

Install

pip install "decimalai>=0.16" strands-agents

Add it to your agent

from decimalai import skillpkg

# Writes .agents/skills and .decimal/skills.toml. Safe to run twice.
result = skillpkg.add("mengto-build-threejs-scroll-worlds")
if result.exit_code:
    raise SystemExit(result.problems)

from decimalai.strands import skills
from strands import Agent
from strands.vended_tools import file_editor, shell

agent = Agent(
    model=model,                # always pass one: the default is Amazon Bedrock
    system_prompt="You are a helpful agent.",
    tools=[file_editor, shell],
    plugins=[skills()],
)
agent("Hello!")

Strands loads an agent's skills once, so build the agent again after decimalai skills update.

No DecimalAI package. The whole skill goes in every prompt, and nothing pins or updates it.

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

Give it to your agent

from pathlib import Path
from strands import Agent

skill = Path(".agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md").read_text()
agent = Agent(model=model, system_prompt=f"You are a helpful agent.\n\n{skill}")
agent("Hello!")

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Strands docs

Adds every file the skill has, pinned so you can update it. Cursor, Copilot, Cline and Warp read this folder by themselves.

pip install "decimalai>=0.15"
decimalai skills add mengto-build-threejs-scroll-worlds

No Python or account. Cursor, Copilot, Cline and Warp read this folder by themselves.

Save it into your project

curl -fsSL --create-dirs -o .agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md https://app.decimal.ai/s/mengto-build-threejs-scroll-worlds/SKILL.md

Copy SKILL.md

Save it in your project, as

.agents/skills/mengto-build-threejs-scroll-worlds/SKILL.md

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Cursor and other editors docs

Fetch the file and give its text to your model as instructions. No account or package needed.

Get the file

curl -fsSL https://app.decimal.ai/s/mengto-build-threejs-scroll-worlds/SKILL.md

Read the summary as JSON

curl -fsSL https://app.decimal.ai/s/mengto-build-threejs-scroll-worlds.json

This gets SKILL.md only. The skill's 5 other files are not included.

Other ways to add it

Other setups docs

You don't need a DecimalAI account.

Version: v1, tested 8/29/2026
Source: mengto/skills ↗
License: MIT

Manage it in DecimalAI

Saved prompts and skills, pinned versions, and traces of this skill in your runs. DecimalAI does not run your agent.

Claude Code reads files from your project, so DecimalAI config does not apply there.

On Claude Agent SDK: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On OpenAI Agents: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On OpenAI SDK: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On Anthropic: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On Gemini API: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On LangChain: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On CrewAI: set DECIMAL_API_KEY and add decimalai.init(crewai=True) to record runs and skill use. The skill version stays in your repo.

On Google ADK: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On Pydantic AI: save the skill to an agent here, set DECIMAL_API_KEY, and the same code uses it with no command to run. Changes you save reach the next run with no redeploy.

On Deep Agents: set DECIMAL_API_KEY and add decimalai.init(langchain=True) to record runs and skill use. The skill version stays in your repo.

On Microsoft Agent Framework: set DECIMAL_API_KEY and add decimalai.init(otel=True, agent_name="support") to record runs and skill use. The skill version stays in your repo.

On Strands: set DECIMAL_API_KEY and add decimalai.init(otel=True, agent_name="support") to record runs and skill use. The skill version stays in your repo.

Editors read files from your project, so DecimalAI config does not apply there.

Anywhere else: record a use with decimalai.log_skill_activation() inside a trace.

Use this skill

SKILL.md

Build Three.js Scroll Worlds

Build one detailed real-time world and use native document scroll as its deterministic conductor. Keep the renderer, scene graph, and spatial continuity alive while camera composition, light, fog, animation, copy, and interaction focus move through authored chapters.

The mechanism is one persistent Three.js world + one normalized reversible scroll state. If removing either makes the experience collapse into stacked sections, this skill applies.

The exact Kage demo proves the quality bar; it is staging, not a mandatory subject or layout. Use its detailed anatomy only when the requested direction benefits from it: references/kage-anatomy.md.

Route the request correctly

Use threejs for a single interactive scene with no scroll-authored journey.
Use scroll-world-storytelling when deciding between real-time 3D, pre-rendered video, and DOM-first storytelling.
Use scroll-scrubbed-visual-sequence for a video or image sequence whose time is scrubbed by scroll.
Use cinematic-scroll-storytelling for DOM-first GSAP/Lenis choreography.
Use this skill when objects must remain truly spatial and interactive while scroll moves through several compositions in one live WebGL world.

Do not disguise a video as Three.js. The public oso95/scroll-world project, reviewed at commit 71cc36d, is a strong reference for intake, scene ledgers, budget gates, mobile-specific composition, config-driven playback, and seam QA, but its renderer is a pre-generated video chain. This skill adopts those structural strengths while retaining real geometry, materials, lighting, raycasting, and camera control.

Read the build references

Before implementing a new world, read:

references/world-bible.md for intake, art direction, scene, material, texture, and interaction ledgers;
references/realtime-architecture.md for the scene graph, conductor, camera, loading, interaction, and lifecycle patterns;
references/quality-and-qa.md before optimization and final verification.

Copy references/scroll-conductor.js when a project needs a portable native-scroll conductor rather than a framework-specific implementation.

Start with the output, not the template

Determine what the world is for: a landing page, portfolio, museum chapter, product explanation, game teaser, editorial essay, data story, or another spatial experience. Do not force every request into a hero-plus-CTA layout.

Discover only what is unknown:

the subject, audience, and ordered story beats;
the art direction and emotional arc;
whether the world is continuous, portal-connected, or composed as adjacent sets;
the target devices and minimum acceptable hardware;
how interactive the world should be beyond scrolling;
available models, textures, brand assets, audio, and copy;
the initial-load and total-download budget.

If the user supplied an approved reference or working scene, inspect it first and carry its real geometry, materials, textures, light ratios, camera values, and motion constants into the ledger. Do not replace measured details with adjectives.

Write the world bible before code

Define one reusable visual grammar:

scale system and unit convention;
silhouette language and level of stylization;
palette and per-chapter accent progression;
material families and wear logic;
key, fill, rim, practical, and emissive light roles;
fog, haze, particles, weather, and post-processing;
typography, DOM composition, and near-plane foreground rules;
motion grammar for camera, ambient objects, and interactions.

Every visible detail must support the same world. Random noise, unrelated texture packs, arbitrary particle effects, and one-off materials create complexity without cohesion.

Author a scene ledger

Use 4–8 chapters for most experiences. Store the full contract as data rather than scattering thresholds through CSS and the render loop:

const chapters = [
  {
    id: "threshold",
    scrollWeight: 1.25,
    copy: {
      eyebrow: "Chapter 01",
      title: "Enter the archive",
      body: "A spatial collection revealed through motion."
    },
    camera: {
      position: [0, 3.8, 13.5],
      target: [0, 2.4, -8],
      fov: 38,
      mobile: { position: [0, 4.8, 18], fov: 46 }
    },
    world: {
      key: 1.0,
      practicals: 0.35,
      fog: 0.018,
      particles: 0.25,
      grade: "cold"
    },
    focus: ["gate", "lantern-left"],
    interactions: ["inspect-gate"],
    assets: ["gate-shell", "stone-1k", "mist-atlas"]
  }
];

For each chapter record the story beat, landmark, camera endpoint, occluders, light/fog state, active animation clips, interaction targets, DOM beat, foreground treatment, asset dependencies, and responsive override. Reject a chapter that differs only by copy.

Build one persistent spatial architecture

Create these systems once:

WebGL canvas
  worldRoot
    environment       sky, terrain, distant silhouettes
    architecture      persistent landmarks and paths
    chapterSets       local props grouped for culling and loading
    interactives      raycast targets and animation state
    atmosphere        fog volumes, particles, weather
    nearPlane3D       optional camera-relative depth accents

DOM above canvas
  semantic chapters  headings, copy, links, media, fallback order
  fixed interface    progress, chapter navigation, controls
  fixed cut-outs      optional alpha foregrounds at viewport edges

Use one renderer and normally one scene. Use layers or render passes when transparency, post-processing, or interaction isolation requires them. Do not instantiate one renderer per chapter or rebuild the world at seams.

Choose a topology deliberately:

Continuous geography: landmarks share one navigable space; strongest presence, highest world-building cost.
Connected sets: rooms or islands are joined by tunnels, gates, fog banks, terrain folds, or darkness; easiest to art-direct and stream.
Layered reveal: the same place changes through time, scale, or state; animate materials, visibility, and light instead of moving to unrelated coordinates.

Hide unavoidable discontinuities behind occlusion, darkness, dense atmosphere, an interior threshold, or a close foreground pass. Never let an object visibly teleport in open space.

Give the world real detail

Geometry

Establish the large silhouette first, then medium forms, then small accents. A dense scatter of tiny props cannot rescue a weak silhouette.
Give architectural edges believable thickness, bevels, joints, caps, reveals, and contact points. Paper-thin roofs and floating props expose the fake immediately.
Use modular kits, instancing, merged static meshes, and LODs instead of duplicating unique geometry.
Break repetition with authored clusters, scale/rotation ranges, decals, and material variation—not unbounded randomness.
Keep collision and raycast proxies simpler than render meshes.

Materials and textures

Use a coherent PBR surface stack where it improves the image:

map	role	common failure
base color	material identity and broad variation	baked highlights fight live lighting
normal	small directional relief	strength too high makes rubber or foil
roughness	controls highlight breakup and age	flat values make every object plastic
AO	contact and crevice grounding	multiplied too heavily makes dirty seams
metalness	separates conductors from dielectrics	gray values everywhere create implausible mud
emissive	practical lights, screens, runes	replaces light but does not illuminate nearby forms
alpha/transmission	foliage, cloth, glass, mist	sorting, overdraw, and depth artifacts

Set color textures to sRGB and keep normal, roughness, AO, metalness, and data textures linear.
Keep texel density consistent within a camera range. Reserve 2K–4K maps for surfaces that genuinely fill the frame; use 512–1K atlases for most props.
Prefer KTX2/Basis compression, Meshopt/Draco geometry compression, atlases, trim sheets, decals, and baked AO/light where appropriate.
Add anisotropy only to shallow-angle surfaces that need it. Setting every texture to the maximum wastes bandwidth.
Make wear causal: exposed edges polish, recesses collect grime, water leaves vertical or pooling traces, and paths compress vegetation. Uniform grunge reads as a filter.
For foliage, use alpha-tested cards or instanced low-poly clusters before transparent blended planes. Transparency overdraw is often more expensive than geometry.

Lighting and atmosphere

Start with one authored key direction, a restrained environment/fill, and practical emissive sources. Add lights only when they create a visible relationship.
Bake static bounce and contact where possible. Keep shadow-casting dynamic lights to roughly 1–2 on mobile and 2–4 on desktop unless profiling proves more.
Use fog and haze to separate depth planes, not merely darken the scene. Keep the focal landmark inside the contrast window.
Match emissive geometry, glow sprites, actual light intensity, and bloom threshold so lamps feel luminous without bleaching the frame.
Use post-processing after the unprocessed frame is composed. Bloom, vignette, grain, chromatic separation, and depth of field must not hide weak materials or framing.

Record the detailed surface and light plan in the ledgers from references/world-bible.md.

Map native scroll to deterministic state

Measure stable section anchors only after fonts and critical media settle. Convert scrollY into a fractional chapter value such as 2.35.

Keep separate values:

rig.target = progressFromScroll(scrollY);       // exact reproducible story state
rig.smooth = reduceMotion
  ? rig.target
  : damp(rig.smooth, rig.target, 5.2, dt);      // cinematic render state

Use exact progress for navigation, URLs, accessibility, foreground ownership, and interaction gating. Use smoothed progress for camera and visual interpolation only. The same scroll position must recreate the same state forward, backward, after a fast jump, and after reload.

Use references/scroll-conductor.js as the baseline implementation. Keep native scroll as the source of truth; never integrate wheel delta into story position.

Author the camera as cinematography

Compose every chapter endpoint at desktop and mobile before interpolating.

Store position, target, FOV, optional roll, lens shift, and responsive overrides.
Use Catmull–Rom curves for broad continuous travel and segment interpolation for deliberate turns.
Inspect the curve for wall penetration, ground clipping, target flips, speed spikes, and accidental close passes.
Parameterize by chapter progress, not raw curve arc length, so story pacing remains intentional.
Use scrollWeight to give important views more dwell; do not distort the camera path just to slow a section.
Add restrained pointer parallax only after the base composition works. Clamp it and blend it out near precise transitions.
On tall viewports, author a pullback/FOV override instead of accepting an arbitrary center crop.

Camera motion must expose new spatial relationships: approach, reveal, passage, scale change, inspection, horizon, departure. Six dolly-ins aimed at the same center are not six scenes.

Interpolate world state from the same conductor

Resolve the adjacent chapters once per frame and interpolate their declared values:

const { a, b, t } = segmentState(rig.smooth);
moon.scale.setScalar(lerp(a.world.moon, b.world.moon, t));
key.intensity = lerp(a.world.key, b.world.key, t);
scene.fog.density = lerp(a.world.fog, b.world.fog, t);

Prefer continuous physical change: occlusion, material blend, animation mixer weight, light, fog, scale, and transform. When swapping assets, crossfade only within an occluded or atmospherically dense interval and keep both states loaded until the transition completes.

Add interactions without fighting scroll

Scroll owns macro movement. Pointer, touch, and keyboard interactions own local response.

Use raycasting against named proxy meshes, not every decorative triangle.
Define idle, hover, focused, active, and unavailable states for each interactive object.
Let hover warm light, shift material response, reveal a label, or trigger a short animation; do not move the scroll camera away from its path.
Pause or soften pointer parallax while the user drags a control or activates a hotspot.
Gate interactions by exact chapter visibility and depth; hidden objects must not capture events.
Mirror every essential hotspot with a DOM button or link in document order. Give it a visible focus state and synchronize its state with the 3D object.
On touch, use tap-to-focus then tap-to-activate when accidental activation would be costly. Keep hit targets at least 44 CSS px in the DOM proxy.
Make decorative particle trails and cursor effects optional, camera-relative where appropriate, and disabled for coarse pointers or reduced motion.

Write the interaction matrix before implementation; use the pattern in references/world-bible.md.

Keep the DOM story semantic

Keep headings, body copy, links, controls, and the footer in real HTML above or beside the canvas. The 3D world creates place; the DOM carries meaning and fallback order.

Reveal display headings by word only when it supports the rhythm; preserve the complete accessible label.
Reveal eyebrow, heading, body, media, proof, and CTA as independent beats.
Keep the reading block stable while the camera makes its largest move.
Use optional fixed alpha cut-outs as near-plane scenery, not as a requirement of every design. When used, keep them fully opaque while active and fade/blur them out after ownership changes.

Load progressively

Make the first authored frame complete before loading the entire journey.

Inline or preload the critical shell, camera, first-scene geometry, smallest environment, and fallback poster.
Load the next chapter before it can enter; prefetch one or two chapters ahead.
Decode textures off the main interaction path and reveal a chapter only when its required assets are ready.
Keep a designed poster and semantic DOM visible if WebGL, a model, or a texture fails.
Report meaningful progress by required asset weight or completed groups, not an arbitrary timer.

Do not hide a 40 MB world behind a decorative percentage. Record the load plan in the asset ledger.

Hold a measurable performance budget

Start with these budgets, then profile on the actual target device:

budget	mobile target	desktop target
DPR cap	1.25–1.5	1.5–2
visible triangles	150k–300k	500k–1.2m
draw calls	50–90	90–160
shadowed lights	1–2	2–4
simultaneously blended full-screen layers	2	3
critical initial transfer	3–6 MB	5–10 MB
steady frame time	≤16.7 ms ideal, ≤25 ms fallback	≤16.7 ms

These are starting envelopes, not success claims. Measure CPU, GPU, texture memory, shader compilation, long tasks, and first-interactive frame.

Cap dt near 1/30 second after stalls.
Pause on document.hidden and when the experience is not visible.
Use a quality governor that lowers DPR and optional effects before deleting authored landmarks.
Render secondary cameras, reflections, cloth simulations, and expensive particles only while relevant.
Dispose geometries, materials, textures, render targets, observers, events, animation mixers, and RAF during teardown.

Respect motion, access, and failure

Preserve native reversible scroll; do not trap the wheel or force a custom scrollbar.
For reduced motion, snap camera/state to the nearest composed chapter, stop ambient loops and stagger/blur, and retain the complete ordered DOM story.
Keep a real heading hierarchy, landmarks, links, form controls, visible focus, and reachable footer.
Provide a composed poster or chapter stills when WebGL is unavailable.
Maintain contrast when the world changes behind copy; use authored scrims or local contrast management, not an opaque blanket over the whole scene.

Verify the whole journey

Follow references/quality-and-qa.md. At minimum verify:

every camera endpoint at 1440×900, 768×1024, and 390×844;
slow, fast, reverse, scrollbar-drag, anchor navigation, reload-at-depth, and resize between chapters;
material scale, UVs, texture color space, roughness response, contact grounding, shadow stability, fog depth, and post-processing;
hover, focus, tap, activation, chapter gating, and interaction recovery after scrolling away and back;
first-load poster, progressive chapter loading, failed asset, context loss, hidden-tab resume, and teardown;
reduced motion, WebGL fallback, keyboard order, readable DOM, reachable footer, and a clean console;
target-device frame time, draw calls, triangles, texture memory, transfer size, and shader compilation.

Use the Codex browser for visual and interaction verification. Build/DOM checks are not visual proof.

Deliver

Return:

the world bible and art-direction constants;
chapter, camera, material/texture, interaction, asset, and performance ledgers;
a layer and scene-graph map;
the working experience and local runtime assets;
desktop, tablet, and mobile evidence from representative chapters;
measured performance and loading results;
fallback/reduced-motion evidence;
remaining visual, interaction, or device gaps stated plainly.

Keep the bundled Kage demo unchanged unless the user explicitly asks to revise the reference itself.

When it activates

Automatically, when a task matches. Use for 3D scrollytelling, scroll-driven WebGL worlds, camera journeys, interactive portfolios, product stories, exhibitions, explainers, game or film microsites, spatial narratives, and multi-scene websites where native scroll should travel through one continuous place. Not limited to landing

About this skill

Automatic

Frontend / UI pack

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

View 15 scanner checks

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

mengto / skills

agent-skills/web-design/build-threejs-scroll-worlds/SKILL.md

View on GitHub ↗

License MITImported 8/10/2026Version v1Bundled 5 files

Illustrative Want to see what mengto/build-threejs-scroll-worlds would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Other skills for the same job

Published skills whose descriptions closely match this one, with their headline benchmark lift.

mengto/scroll-world-storytellingTurn an article, case study, brand narrative, product journey, or long-form story into a cinematic scroll-driven landing page using one of three renderers: scrubbed video, a real-time Three.js world, or semantic HTML/SVG data and typography. Use when the user asks for a scroll world, fly-through landing page, article-to-website transformation, animated planet, data scrollytelling, video-scrubbed page, connected visual journey, or story-led alternative to ordinary stacked sections.Close match+41 pts pass-rate lift

Related skills

Skills that do similar work, with their headline benchmark lift.

From the same source and nearby in the catalogue.

Ready to supercharge your AI agents?

SkillScore

Safety

What it changes

Track record

Use this skill

Build Three.js Scroll Worlds

Route the request correctly

Read the build references

Start with the output, not the template

Write the world bible before code

Author a scene ledger

Build one persistent spatial architecture

Give the world real detail

Geometry

Materials and textures

Lighting and atmosphere

Map native scroll to deterministic state

Author the camera as cinematography

Interpolate world state from the same conductor

Add interactions without fighting scroll

Keep the DOM story semantic

Load progressively

Hold a measurable performance budget

Respect motion, access, and failure

Verify the whole journey

Deliver

When it activates

SkillSafety

Source

Other skills for the same job

Related skills

More in the registry