Ready to supercharge your AI agents?

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

Get Started Free →

Multi-Agent Workshop · leoyeai | DecimalAI

Agent Toolingby leoyeai

Multi-Agent Workshop

七阶段（含阶段0任务澄清）多角色工作坊；角色种类与人数均由任务决定。触发词："多角色研讨"、"需求工作坊"、"需求评审"、"圆桌"、"定方案再执行"。

SkillScore

SkillScore64/100

Test results 64

Safety

3/3 safety checks passed

Code scan
Intent
Content

What it changes

14%64%of test cases passed +50 pts

Without skill3 of 22

With skill14 of 22

11 fixed3 passed either way8 still failing0 broke

Measured on gemini-3.6-flash8/10/2026v119 comparable

Track record

Agent runs: 0; All time

See one task, with and without the skill

One test taskWe are creating a single-deliverable script generation plan with only two execution steps. We want to use plan-lite.md instead of the full plan template. What explicit metadata tag must be included in plan.md?

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 leoyeai-multi-agent-workshop
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("leoyeai-multi-agent-workshop")
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 leoyeai-multi-agent-workshop

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/leoyeai-multi-agent-workshop/SKILL.md https://app.decimal.ai/s/leoyeai-multi-agent-workshop/SKILL.md

Copy SKILL.md

Save it in your project, as

.claude/skills/leoyeai-multi-agent-workshop/SKILL.md

This gets SKILL.md only. The skill's 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from agents import Agent, Runner

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from openai import OpenAI

client = OpenAI()
skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
import anthropic

client = anthropic.Anthropic()
skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/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/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from langchain.agents import create_agent

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from pydantic_ai import Agent

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from deepagents import create_deep_agent

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

import asyncio
from pathlib import Path
from agent_framework import Agent

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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("leoyeai-multi-agent-workshop")
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/leoyeai-multi-agent-workshop/SKILL.md

Give it to your agent

from pathlib import Path
from strands import Agent

skill = Path(".agents/skills/leoyeai-multi-agent-workshop/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 15 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 leoyeai-multi-agent-workshop

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/leoyeai-multi-agent-workshop/SKILL.md https://app.decimal.ai/s/leoyeai-multi-agent-workshop/SKILL.md

Copy SKILL.md

Save it in your project, as

.agents/skills/leoyeai-multi-agent-workshop/SKILL.md

This gets SKILL.md only. The skill's 15 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/leoyeai-multi-agent-workshop/SKILL.md

Read the summary as JSON

curl -fsSL https://app.decimal.ai/s/leoyeai-multi-agent-workshop.json

This gets SKILL.md only. The skill's 15 other files are not included.

Other ways to add it

Other setups docs

You don't need a DecimalAI account.

Version: v1, tested 8/10/2026
Source: LeoYeAI/openclaw-master-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

Multi-Agent Workshop

在单会话内由主 Agent 担任导演，按 **7 个阶段（含阶段 0）**推进；每阶段产物写入 workshops/<session_id>/，角色发言建议用 subagent（一次一个角色、只注入该角色立场），避免上下文混乱。

所有阶段跳转必须经 orchestrator.py（Plan B）：门禁以代码强制，不依赖 Markdown 约定。导演每次推进阶段时执行 orchestrator.py advance <sid>；跳转前检查 orchestrator.py gate <sid>。

何时启用

用户要：多人视角讨论同一需求、互评、输出可执行方案、确认后再执行。

核心原则：角色由任务决定，不是固定班子

禁止不经 阶段 0 就默认把「如何产出高质量 X」当成 必须交完整成稿（或相反）；须先锁定 交付物类型（方案/SOP、成稿、大纲+样例、混合等）。
禁止不经阶段 2 就默认使用「运营 / 产品 / 技术」；三者仅是通用软件需求的参考模板，不是每题的答案。
阶段 2 必须先回答：「要完成这个任务，最少需要哪几种相互制约的专业视角？」再列角色。
角色人数没有固定值（不是必须 2 个、3 个或 5 个）：完全由任务决定——有几个不可缺少且彼此能形成张力的视角，就设几个角色；宁可少也不要凑数。多一个角色就多一轮 token 与协调成本。
下文的「2～5」「调用量级表」仅是常见规模参考，不是规范；唯一标准是阶段 2 的选角理由是否成立。
在 state.md 必须写清：任务类型标签 + 为何选这些角色（及为何是这个人数量）（各一行），便于复盘。

任务类型 → 角色组合（示例，可增删）

任务类型	常见角色视角（示例）	说明
通用产品需求 / 功能评审	运营、产品、技术	参考 `references/sample-roles/`
短视频 / 直播 / 投放素材	编导或内容、法务或合规、剪辑或制作	重传播与合规，未必需要「技术」
公关 / 舆情 / 对外声明	公关、法务、业务事实负责人	避免只有运营
数据 / 指标 / 实验	数据/分析、产品、工程	「运营」可换成增长但职责要写清
纯技术债 / 架构改造	技术、SRE/运维、产品（仅定范围）	可不要运营
采购 / 商务 / 合同	商务、法务、财务或使用方	与产研三角无关
法律合规审查	法务、业务、执行对接人	不要假装「技术」能替代法务立场
市场研报 / 行业研究	市场情报、分析或产品视角、事实/合规把关	需可引用事实时优先安排一次集中检索；检索失败须在 `state` 与正文披露

导演动作：从上表得到启发即可，不要机械套用；若用户任务跨界，可合并角色（一人兼两视角但分两轮说）或拆两次工作坊。

目录与产物（相对 workspace 根）

产物	路径
工作坊根目录	`workshops/<session_id>/`
状态 + 阶段 + 摘要	`workshops/<session_id>/state.md`
完整发言（可选）	`workshops/<session_id>/transcript.md`
可执行方案	`workshops/<session_id>/plan.md`
交付正文（推荐）	`workshops/<session_id>/deliverables/*`（报告、脚本包、设计稿等）

deliverables/：阶段 6 或同步写入的对用户可见成果默认放此目录，避免与 state.md 混写；路径写进 plan.md 执行清单或验收表。

敏感信息：workshops/ 可能含未公开需求；对外 git 请忽略或脱敏，见 workshops/README.md。

session_id 建议：YYYY-MM-DD_主题简写。用户未命名时由导演生成。

在 state.md 的「当前阶段」中统一使用下列 phase key（便于检索）：

阶段	phase key
0 任务澄清	`phase_0_intake`
1 任务理解	`phase_1_understanding`
2 角色拆解	`phase_2_roles`
3 角色创建	`phase_3_role_cards`
4 任务讨论	`phase_4_discussion`
5 plan 起草	`phase_5_plan`
5 末尾待批	`awaiting_approval`（`plan.md` 已落盘、用户未确认前）
6 plan 执行	`phase_6_execution`

说明：awaiting_approval 仍属阶段 5 的末尾，不是第 7 阶段；用户确认后进入 phase_6_execution。

模板与骨架（复制起盘）

用途	路径
`state.md` 结构	`templates/state.example.md`
`transcript.md` 结构	`templates/transcript.example.md`
自建角色卡	`references/role-card-skeleton.md`
从 JD 构建角色卡	`references/role-card-from-jd.md`
阶段 1 导演迫问（可选）	`references/director-forcing-questions.md`
`plan.md`（全量）	`templates/plan-output.md`
`plan.md`（轻量）	`templates/plan-lite.md`（见下文「plan 模板选择」）
Orchestrator（唯一阶段管理入口）	`scripts/orchestrator.py`（阶段机 + 门禁 + SQLite；硬性阻塞不满足前置条件的跳转）
阶段 6 执行脚本生成（可选）	`workshops/scripts/generate_execution.py`（`--preset` 或 `--from-plan` 读 `plan.md` 的 `execution_*`）；说明见 `workshops/scripts/README.md`
JD→角色卡草稿生成	`scripts/jd_to_role_card.py`（`--role <名> --industry <行业> [--task <任务>]`）
各阶段检查清单（给人/LLM 看）	`scripts/phases/00-…06-*.md`
关闭 subagent 并行（改 `openclaw.json`）	`scripts/openclaw-subagents-parallel.sh`（`off` = `maxConcurrent`→1；需 `python3`；改前自动备份）

Orchestrator CLI（唯一阶段管理入口）

scripts/orchestrator.py：Python 3.8+，无第三方依赖，状态存 SQLite（data/orchestrator.db）。所有阶段跳转必须经此工具，门禁以代码强制。

子命令	说明
`orchestrator.py init <sid> [-d TYPE]`	创建 session + workshops 目录（可选指定 deliverable_type）
`orchestrator.py status <sid>`	当前状态 + 下一门禁是否通过
`orchestrator.py advance <sid>`	推进到下一阶段（检查门禁）
`orchestrator.py set-phase <sid> <phase> [--force]`	跳转阶段（检查门禁；`--force` 可强行跳过）
`orchestrator.py gate <sid> [phase]`	查看门禁状态
`orchestrator.py set <sid> <field> <value>`	设置字段（deliverable_type、raw_requirement、approved_at 等）
`orchestrator.py set-role <sid> <name> [-r ...] [-t ...] [--card-path ...]`	添加/更新角色
`orchestrator.py add-message <sid> --phase ... --role ...`	记录讨论消息
`orchestrator.py validate <sid>`	全量校验
`orchestrator.py history <sid>`	阶段变更历史
`orchestrator.py import <sid>`	从现有 `workshops/<sid>/state.md` 导入
`orchestrator.py sync <sid>`	DB 状态回写 `state.md`（phase key + Session 状态）
`orchestrator.py cleanup`	清理 OpenClaw 残留 subagent
`orchestrator.py list`	列出所有 session

门禁规则（代码强制，非建议）：

目标阶段	前置条件
→ 阶段 1	阶段 0 填了 deliverable_type + success_criteria
→ 阶段 2	阶段 1 填了 raw_requirement
→ 阶段 3	阶段 2 填了 task_type_tag + role_rationale + ≥1 角色
→ 阶段 4	阶段 3 每个角色有 card_path 或 card_inline
→ 阶段 5	阶段 4 有 ≥1 条讨论消息
→ awaiting	plan.md 存在或 plan_version 已设
→ 阶段 6	当前为 awaiting_approval 且 approved_at 已设

示例（完整起盘到推进流程）：

cd /path/to/workspace
SID="2026-03-22_my-topic"
ORC="python3 skills/multi-agent-workshop/scripts/orchestrator.py"

# 1. 起盘（创建 session + workshops 目录 + 模板文件）
$ORC init $SID -d methodology

# 2. 阶段 0：填交付物类型、成功标准
$ORC set $SID success_criteria "产出可复用的写作方法论 SOP"
$ORC advance $SID                    # → phase_1_understanding（门禁检查）

# 3. 阶段 1：填原始需求
$ORC set $SID raw_requirement "如何产出高质量 AI 观点文章"
$ORC advance $SID                    # → phase_2_roles

# 4. 阶段 2：填任务类型、选角理由、添加角色
$ORC set $SID task_type_tag "内容创作"
$ORC set $SID role_rationale "需要信息捕捉、观点提炼两种张力"
$ORC set-role $SID 趋势猎手 -r "扫描热点" -t "热点vs深度"
$ORC set-role $SID 观点锻造师 -r "提炼角度" -t "差异化vs可论证"
$ORC advance $SID                    # → phase_3_role_cards

# 5. 随时检查状态 / 门禁 / 历史
$ORC status $SID
$ORC gate $SID
$ORC history $SID

防走样（导演必读）

不标 director_monologue / 压缩就禁止单口吻写多角色全文。
省略 subagent 或迫问须在 state.md + plan.md 风险同时披露。
跳过迫问清单须写具体原因（禁止仅用「已内化」「略」）。

完整规则（独白/压缩叠加、何时不可压缩、迫问跳过格式）→ references/anti-drift.md

导演补充规则

以下为低频参考，完整内容 → references/director-supplementary.md

主题	要点	详见
用户确认	plan 勾选、自然语言「确认执行」、或子集确认均可；模糊表述不算	`director-supplementary.md` §用户确认
轮次成本	2 角色≈4 调用，3≈6；用户要快可压缩但须标注	同上 §轮次与成本
事实检索	阶段 4 前至多 1-2 次集中检索；失败须声明，禁止捏造	同上 §事实检索
会话续跑	同 SID 续写须先读 state/plan；默认新 SID	同上 §会话续跑
subagent 收口	每跳有结束说明或导演补录；session 结束写 state.md	`references/subagent-closure.md`
OpenClaw 集成	阶段 6 遵循 `AGENTS.md` 路由 + `TOOLS.md`	`director-supplementary.md` §OpenClaw
gstack 参考	只读克隆 `workspace/reference/gstack/`	`docs/gstack-learning-notes.md`
辩论协议	提前收束、冲突处理、语言约定	`references/debate-protocol.md`

全局约束（全程有效）

阶段 0～5：以分析、文档、讨论为主；不擅自执行对外发送、改生产、支付等离开本机的操作。
阶段 6：仅当用户在对话中明确确认 plan.md（或确认修改后的版本）后启动；高风险步骤先复述再执行。

阶段 0：任务澄清（phase_0_intake）

目的：在选角与多轮讨论前，锁定 交付物类型 与 非目标，避免典型走样：用户要「可复用的写作方法论」，系统却直接交付「一篇完整文章」（或相反）。

何时必须做：用户表述模糊（如「如何产出高质量文章」「帮我做好内容」）且未明确 交付形态 时；默认新 session 从本阶段开始（state.md 模板默认 phase_0_intake）。

动作：

在 state.md「阶段 0」勾选或填写 期望交付物类型（见 templates/state.example.md）：methodology / artifact / outline_sample / review / mixed（混合须写清边界与顺序）。
写 非目标（例：本 session 不默认交付 3000 字成稿；或不只给空洞框架而不给可执行步骤）。
写 成功标准 + 用户原话摘要 + 导演对齐说明（有歧义须追问，禁止单方面猜交付物）。
若用户改口改变交付物类型：回到本阶段更新 state.md，或进入阶段 5+ 时 递增 plan_version 并在风险中披露。

完成条件：阶段 0 字段齐全 → 执行 orchestrator.py set <sid> deliverable_type <type> + set <sid> success_criteria "..." → orchestrator.py advance <sid>（门禁检查 deliverable_type + success_criteria）。

阶段 1：任务理解（phase_1_understanding）

目的：在 交付物类型已锁定 的前提下，把「用户到底要什么细节」说清楚，避免后面角色空转。

动作：

从用户输入提取：业务目标、用户/客户、时间约束、成功标准；缺失则写假设并标「待用户确认」。不得在本阶段悄悄改阶段 0 的交付物类型；若需改则回到阶段 0 或走 plan 变更流程。
推荐：按 references/director-forcing-questions.md 择问（不必六问全问），把答案并入上文假设或「待澄清」；未使用清单时须满足 references/anti-drift.md「迫问清单跳过格式」。用户要求快进时可同步启用 压缩模式。
在 state.md 写入 原始需求（可含原文摘要 + 导演理解的一段话）。

完成条件：state.md 中「原始需求 + 约束与假设」齐全 → 执行 orchestrator.py set <sid> raw_requirement "..." → orchestrator.py advance <sid>（门禁检查 raw_requirement）。

阶段 2：角色拆解（phase_2_roles）

目的：为当前这一条任务定制视角组合；不是默认固定三人。

动作：

在 state.md 写 任务类型标签（如：短视频、功能开发、数据需求、公关声明）。
自问：哪些视角之间会有真实张力（例如传播 vs 合规、范围 vs 工期）？只邀请这些视角进组。
若选用 运营 / 产品 / 技术：必须在 state.md 写一句 「为何本任务适用产研三角」；否则视为阶段 2 未完成。
若任务与产研无关，不得硬塞 ops/product/tech；应改用阶段 3 的 session 角色卡 定义新角色。
写入 角色列表：角色名 + 一句话职责 + 在本任务中最关心的冲突点。

完成条件：任务类型 + 选角理由（含为何是 N 个角色、能否再少）+ 角色列表齐全 → 执行 orchestrator.py set <sid> task_type_tag "..." + set <sid> role_rationale "..." + set-role <sid> <name> -r ... -t ...（每个角色） → orchestrator.py advance <sid>（门禁检查 task_type_tag + role_rationale + ≥1 角色）。若阶段 1 已用迫问清单，选角应与迫问暴露的张力一致。

阶段 3：角色创建（phase_3_role_cards）

目的：每个角色有可执行的立场说明（给 subagent 用），且与当前任务绑定。

动作：

对阶段 2 确定的每个角色，导演按下表判断角色卡来源，然后执行对应路径：

条件	路径	做什么
角色与 `sample-roles/`（ops/product/tech）语义一致，且导演能准确写出该角色在本任务中的立场边界	复用样例	复制 `references/sample-roles/*.md`，在开头加「本任务中你的侧重点」段
角色是通用职能（如项目经理、设计师）但不在 sample-roles 中，导演对该岗位职责边界有把握	直接新建	按 `references/role-card-skeleton.md` 手写
以下任一为真则必须查 JD ↓	JD 构建	搜索 JD → 提炼 → 写卡（见下方流程）
① 角色属于垂直/专业领域（如合规法务、供应链、临床研究、财税）——导演不确定该岗位真实的职责边界和禁区
② 角色名在团队日常中不常见或导演从未写过该角色的卡
③ 阶段 2 选角时导演犹豫过该角色"到底管什么、不管什么"

JD 构建流程（满足上表"必须查 JD"条件时执行）：

搜索：执行 python3 scripts/jd_to_role_card.py --role "<角色名>" --industry "<行业>" --task "<本次任务>" 生成草稿；或在对话中搜索 "<角色名>" <行业> 岗位职责任职要求 BOSS直聘 OR 猎聘。
提炼：从 JD 片段中提取 职责→立场、任职要求→发言要求、岗位边界→禁止项（映射规则见 references/role-card-from-jd.md）。
绑定：首行写明 针对任务：〈…〉，将通用 JD 画像裁剪为本任务的具体侧重点。
落盘：写入 workshops/<session_id>/roles_<slug>.md。

不查 JD 也可以，但要说明原因：若导演认为自己对该角色足够了解而跳过 JD 查询，须在 state.md 角色列表旁写一句 「未查 JD，理由：…」（如"该角色与 sample-roles/ops 语义一致"）。禁止无声跳过。

完成条件：每个参与讨论的角色均有可注入文本 → 执行 orchestrator.py set-role <sid> <name> --card-path <path> 或 --card-inline "..." → orchestrator.py advance <sid>（门禁检查每个角色有 card）。

阶段 4：任务讨论（phase_4_discussion）

目的：先各抒己见，再互评，形成可写进 plan 的共识与分歧。

动作：

4a 观点轮（每角色一次或按 `max_rounds` 控制）

subagent 任务模板：

你是【角色名】，立场与输出格式必须严格遵守：
（粘贴对应 session 角色卡全文或 references/sample-roles/*.md）

共同上下文：
（粘贴 state.md 中的「原始需求」+ 已有发言摘要）

本轮任务：
- 阐述你对需求的理解（目标、范围、指标）
- 明确你与其他角色可能冲突的假设
- 提出至少 2 条可验证的待澄清问题
输出：中文 Markdown，小标题分段。
末尾必须附：**### 本子 agent 结束说明**（见 `references/subagent-closure.md` 模板）。

每角色结束后，将**≤400 字摘要**写入 state.md「轮次摘要」；若模型未附结束说明，导演补录。

4b 互评轮

subagent 任务改为：

回应其他角色理解中与你不一致之处；
指出对方具体不足（附「若成立则影响」）；
你的让步条件与底线。末尾必须附：### 本子 agent 结束说明（同上）。

摘要同样写入 state.md；无结束说明则导演补录。

Token 控制：默认观点轮 1 轮 + 互评 1 轮；若角色 >3 或用户要求缩短，导演可在 state.md 说明「压缩轮次」并在 plan「风险」中记录。

完成条件：各角色完成观点 + 互评（或导演提前收束并说明）→ 每轮用 orchestrator.py add-message <sid> --phase phase_4_discussion --role <role> --summary "..." 记录 → orchestrator.py advance <sid>（门禁检查 ≥1 条讨论消息）。

阶段 5：plan 建立（phase_5_plan）

目的：把讨论收敛为单一可执行文档，并等人拍板。

plan 模板选择

条件	使用模板	必填标记
多里程碑、执行清单 ≥3 项、需完整风险矩阵	`templates/plan-output.md`	默认
单一交付物（如一篇报告、一个视频包）、清单 ≤2 项	`templates/plan-lite.md`	`plan.md` 内写 `plan_template: lite`，且 §风险说明相对全量模板缺了哪些节

全量与 lite 不可混用结构不声明；升级 scope 时可将 lite 升格为全量并递增 plan_version。

动作：

主 Agent（导演）阅读 state.md 摘要，必要时回看 transcript.md。
按所选模板写 plan.md（全量：共识、指标、里程碑、执行清单、风险、plan_version、CHANGELOG 约定见 plan-output.md；lite 见 plan-lite.md）。
汇总待澄清项：已解决的写入共识表；未决写入「未决项」或 plan 风险。
执行 orchestrator.py set <sid> plan_version 1 → orchestrator.py advance <sid>（门禁检查 plan.md 存在）→ 自动进入 awaiting_approval。
在对话中明确请用户确认或修改 plan.md。

完成条件：plan.md 已落盘 → 用户确认后执行 orchestrator.py set <sid> approved_at "YYYY-MM-DDTHH:MM" → orchestrator.py advance <sid>（门禁要求 awaiting_approval + approved_at 已设）→ 进入 phase_6_execution。

阶段 6：plan 执行（phase_6_execution）

目的：按清单调用 工作区 skills/（与 Cursor @workspace/skills 同目录、相对工作区根）内的技能及工具，并留痕。

路径约定：<技能名> 与 AGENTS.md 路由表一致；每个原子步骤打开并遵循 skills/<技能名>/SKILL.md，不得凭记忆编造脚本路径或技能名。

本机与环境：阶段 6 执行须对照工作区根 TOOLS.md（与 Cursor @workspace/TOOLS.md 同文件）——浏览器、Obsidian、cron、飞书附件、基础库路径、生图/生视频端到端流程（提示词技能 vs 出图/海螺脚本顺序）等；不要求每次通读全文，但任务涉及哪一节就必须读哪一节。细则与检查清单见 scripts/phases/06-phase_6_execution.md §0 表格。

阶段 6 执行清单（给人/导演逐步核对）：scripts/phases/06-phase_6_execution.md（含「AGENTS.md + TOOLS.md + skills/」完整流程）。

技能搭配（阶段 6：先扫描装配，再执行）：

用户一旦确认方案（approved_at 已写入），阶段 6 分两节拍（详见 scripts/phases/06-phase_6_execution.md §0）：

节拍 6a — 技能扫描与装配（必先做）：对照 plan.md 执行清单逐步读 AGENTS.md → 按任务打开 TOOLS.md 对应小节 → 为每步打开并核对相关 skills/<name>/SKILL.md，再决定「用哪把锤子」；未完成 6a 前，不得调用计划外 API、不得用临时方案冒充交付物。
节拍 6b — 按 SKILL 执行：严格按各 SKILL.md 执行，并与 TOOLS.md 本机约定对齐；留痕。

完整装配算法见 references/skill-composition.md。

skill-pipeline.md：可选落盘（多步/多技能/要审计时强烈建议）；步骤少时可在 execution-log.md 用几句话记录「已扫 AGENTS、选用/排除谁」即可 —— 省略的是文档长度，不是 6a 扫描本身。
可选机器化：若流水线某步对应本机可跑工具链（如 OpenRouter 出图复用 xhs-card-generator 客户端），可在 plan.md 写 execution_preset / execution_prompts_file（见 templates/plan-lite.md §7）后运行 workshops/scripts/generate_execution.py --workshop workshops/<session_id> --from-plan；或直接 --preset <名>。在该会话目录下生成 scripts/run_*.py（详见 workshops/scripts/README.md）。
红线：不虚构技能名；互斥以 AGENTS.md 为准；纯人工步骤标 human_design，不得冒充完成。

动作：

仅当用户已确认方案（可在 plan.md 用户确认章节勾选并填确认时间）。
将执行清单逐项落实；每完成一项更新 plan.md 状态列或写 execution-log.md（若使用了 skill-pipeline.md，序号与之对齐）。
每一次 subagent / 独立工具链任务结束时，须满足 references/subagent-closure.md 的「单次结束」格式（产出摘要、完成度、对导演的提示）；若模型未输出，导演补录到 transcript.md 或 state.md。
若执行中发现需改 scope，暂停，回到阶段 5 更新 plan.md 并再次确认。

完成条件：清单完成或用户主动叫停 → 执行 orchestrator.py set <sid> session_status completed（或 stopped / cancelled）+ set <sid> ended_at "..." → orchestrator.py sync <sid>（回写 state.md）。并填写 state.md 「结束与归档」（见 templates/state.example.md）。

质量检查

□ 阶段 0：`state.md` 含 **期望交付物类型** + **非目标** + **成功标准** + 与用户原话对齐（避免「要方案却给成稿」）
□ state.md 含任务类型标签 + 选角理由（含人数为何是 N）；非产研任务未滥用 ops/product/tech；未为凑人数硬加角色
□ 阶段 1：已用迫问或写明 **「未使用 director-forcing-questions.md，原因：…」**（禁止仅写「已内化」）
□ 若有交付文件，路径在 `deliverables/` 且 `plan` 可验收
□ 选用 `plan-lite` 时已标 `plan_template: lite` 且风险说明与全量模板的差异
□ 阶段 2 角色列表与任务匹配；阶段 3 每角色有任务绑定下的角色卡（门禁已验证「针对任务：…」；骨架见 references/role-card-skeleton.md）
□ phase key 含 awaiting_approval 语义时与正文一致，无「口头阶段与文末矛盾」
□ 阶段 4：每角色独立调用或已标注 director_monologue / 压缩模式
□ plan.md 含 `plan_version`；**全量**模板须含 Owner、依赖、验收标准；**lite** 至少含交付物路径与验收（见 `plan-lite.md`）；阶段 6 前满足「用户确认」一节任一条件
□ 未经确认不进入阶段 6；workshops 敏感内容已考虑是否进 .gitignore
□ 阶段 6：已做 **6a 扫描装配** 再 **6b 执行**（见 `06-phase_6_execution.md` §0）；技能选用对照 **`AGENTS.md` + `TOOLS.md`（任务相关节）+ `skills/<name>/SKILL.md`**；多步已按需落盘 `skill-pipeline.md` 或在 `execution-log.md` 留装配摘要（见 `skill-composition.md`）
□ 子 agent：每跳有结束说明或导演已补录（见 `references/subagent-closure.md`）；session 结束时 `state.md` 含 **Session 状态** + **结束与归档**

About this skill

Automatic

AI Engineer pack

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/2/2026

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

LeoYeAI / openclaw-master-skills

skills/multi-agent-workshop/SKILL.md

View on GitHub ↗

License MITImported 8/2/2026Version v1Bundled 15 files · scripts

Illustrative Want to see what leoyeai/multi-agent-workshop would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

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

Multi-Agent Workshop

何时启用

核心原则：角色由任务决定，不是固定班子

任务类型 → 角色组合（示例，可增删）

目录与产物（相对 workspace 根）

模板与骨架（复制起盘）

Orchestrator CLI（唯一阶段管理入口）

防走样（导演必读）

导演补充规则

全局约束（全程有效）

阶段 0：任务澄清（phase_0_intake）

阶段 1：任务理解（phase_1_understanding）

阶段 2：角色拆解（phase_2_roles）

阶段 3：角色创建（phase_3_role_cards）

阶段 4：任务讨论（phase_4_discussion）

4a 观点轮（每角色一次或按 max_rounds 控制）

4b 互评轮

阶段 5：plan 建立（phase_5_plan）

plan 模板选择

阶段 6：plan 执行（phase_6_execution）

质量检查

SkillSafety

Source

Related skills

More in the registry

4a 观点轮（每角色一次或按 `max_rounds` 控制）