Ready to supercharge your AI agents?

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

Education Data Context · brycewang-stanford | DecimalAI

Databy brycewang-stanford

Education Data Context

Interpretation guidance for Urban Institute Portal datasets. Coded values (-1/-2/-3), year definitions, grade encoding, suppression, licensing, cross-source joins.

Never made a case worse

SkillScore

SkillScore95/100

Test results 91 · Live runs 100

Safety

3/3 safety checks passed

Code scan
Intent
Content

What it changes

59%91%of test cases passed +32 pts

Without skill13 of 22

With skill20 of 22

7 fixed13 passed either way2 still failing0 broke

Measured on gemini-3.6-flash8/9/2026v1

Track record

Agent runs: 37; All time

See one task, with and without the skill

One test taskAn analyst needs deep historical context on IPEDS accounting standard changes (GASB vs FASB) and detailed survey component definitions across all postsecondary surveys. Which specific deep-dive skill should they load?

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 brycewang-stanford-education-data-context
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("brycewang-stanford-education-data-context")
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 brycewang-stanford-education-data-context

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/brycewang-stanford-education-data-context/SKILL.md https://app.decimal.ai/s/brycewang-stanford-education-data-context/SKILL.md

Copy SKILL.md

Save it in your project, as

.claude/skills/brycewang-stanford-education-data-context/SKILL.md

This gets SKILL.md only. The skill's 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from agents import Agent, Runner

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from openai import OpenAI

client = OpenAI()
skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
import anthropic

client = anthropic.Anthropic()
skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/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/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from langchain.agents import create_agent

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from pydantic_ai import Agent

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from deepagents import create_deep_agent

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

import asyncio
from pathlib import Path
from agent_framework import Agent

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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("brycewang-stanford-education-data-context")
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/brycewang-stanford-education-data-context/SKILL.md

Give it to your agent

from pathlib import Path
from strands import Agent

skill = Path(".agents/skills/brycewang-stanford-education-data-context/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 6 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 brycewang-stanford-education-data-context

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/brycewang-stanford-education-data-context/SKILL.md https://app.decimal.ai/s/brycewang-stanford-education-data-context/SKILL.md

Copy SKILL.md

Save it in your project, as

.agents/skills/brycewang-stanford-education-data-context/SKILL.md

This gets SKILL.md only. The skill's 6 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/brycewang-stanford-education-data-context/SKILL.md

Read the summary as JSON

curl -fsSL https://app.decimal.ai/s/brycewang-stanford-education-data-context.json

This gets SKILL.md only. The skill's 6 other files are not included.

Other ways to add it

Other setups docs

You don't need a DecimalAI account.

Version: v1, tested 8/9/2026
Source: brycewang-stanford/Auto-Empirical-Research-Skills ↗
License: NOASSERTION

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

Education Data Context

Data origin, caveats, and interpretation guidance for Urban Institute Education Data Portal datasets. Use when interpreting Portal coded values (-1/-2/-3 missing/not-applicable/suppressed), understanding year definitions (fall vs. academic year), applying correct grade encoding (grade=-1 means Pre-K, not missing), assessing suppression rates, citing data under ODC-By license, or reviewing any Portal data before analysis. Also covers joining identifiers across CCD, IPEDS, CRDC, and other sources, and routes to source-specific deep-dive skills.

This skill provides critical context for interpreting data from the Urban Institute Education Data Portal. Education data has source-specific limitations that can significantly affect analysis validity.

Why Data Context Matters

Source-specific limitations: Each data source (CCD, IPEDS, CRDC, etc.) has unique constraints
Missing values have meaning: Codes like -1, -2, -3 indicate specific conditions, not random missingness
Definitions change over time: Variable definitions, categories, and coding schemes evolve
State comparisons require caution: State-level data often cannot be directly compared
Citation is required: The ODC Attribution License mandates proper citation
Skill provenance matters: Each *-data-source-* skill includes provenance.skill_last_updated in its frontmatter. If this date is more than a few months old, treat the skill's claims about coded values, suppression patterns, and data quality with caution — data sources evolve and skill documentation may have drifted. Consider re-running data-ingest to re-verify.

Data Provenance: The Education Data Portal

All education data currently accessible through this system is obtained from the Urban Institute Education Data Portal (EDP), not directly from original source agencies (NCES, Census Bureau, Department of Education, etc.). The EDP is a curation and standardization layer that:

Renames variables to lowercase (e.g., enrollment not MEMBER)
Re-encodes categoricals as integers (e.g., 1 not "Regular school")
Standardizes missing values using codes -1 (missing), -2 (not applicable), -3 (suppressed)
May subset each source's full variable catalog — not all variables from the original source are necessarily available through the Portal

Each education-data-source-* skill documents what is available through the Portal for that source, including any known gaps relative to the original data collection. When a skill also documents variables or components only available from the original source directly, this is clearly noted.

Note: This provenance applies specifically to the current education data source skills. Future data source skills may access data from other providers with different characteristics.

Reference File Structure

Quick Context (This Skill)

File	Content	When to Read
`./references/ccd-context.md`	K-12 schools/districts caveats	After pulling CCD data
`./references/ipeds-context.md`	College/university caveats	After pulling IPEDS data
`./references/crdc-context.md`	Civil rights data caveats	After pulling CRDC data
`./references/scorecard-context.md`	College Scorecard caveats	After pulling Scorecard data
`./references/edfacts-context.md`	Assessment/graduation caveats	After pulling EDFacts data
`./references/data-relationships.md`	Joining tables, identifiers	When merging datasets

Deep-Dive Source Skills (Comprehensive Documentation)

These skills document both EDP-available data and original source context. Each skill notes when content applies only to the original source (not available through the Portal).

For comprehensive understanding beyond the quick context files above, load the dedicated data source skill:

Data Source	Deep-Dive Skill	Key Deep Topics
CCD	`education-data-source-ccd`	Survey components, EDFacts submission, state variations, historical changes
CRDC	`education-data-source-crdc`	Civil rights legal context, underreporting issues, year-to-year evolution
EDFacts	`education-data-source-edfacts`	ESSA/NCLB context, why states aren't comparable, ACGR methodology
IPEDS	`education-data-source-ipeds`	All 12+ surveys, graduation rate population limits, GASB vs FASB
Scorecard	`education-data-source-scorecard`	IRS earnings methodology, Title IV selection bias, suppression rules
SAIPE	`education-data-source-saipe`	Model-based estimation, no district confidence intervals
FSA	`education-data-source-fsa`	Title IV programs, financial responsibility scores, 90/10 rule
MEPS	`education-data-source-meps`	Superior to FRPL for cross-state poverty comparison
NHGIS	`education-data-source-nhgis`	Census geography links, boundary changes over time
NACUBO	`education-data-source-nacubo`	Endowment study methodology, voluntary participation bias
NCCS	`education-data-source-nccs`	Form 990 data, NTEE codes, private college relevance
EADA	`education-data-source-eada`	Title IX context, not same as compliance data
Campus Safety	`education-data-source-campus-safety`	Clery Act, underreporting, geography definitions
PSEO	`education-data-source-pseo`	LEHD methodology, experimental status, state coverage

When to load deep-dive skills:

Need to understand data collection methodology in detail
Analyzing historical trends and need to know about definition changes
Encountering data quality issues that require deeper investigation
Writing documentation or reports that require precise methodology descriptions

Decision Trees

What data source did I pull from?

What endpoint did you use?
├─ schools/ccd/* → Read ./references/ccd-context.md
│   └─ Need more depth? → Load education-data-source-ccd skill
├─ school-districts/* → Read ./references/ccd-context.md
│   └─ Need more depth? → Load education-data-source-ccd skill
├─ schools/crdc/* → Read ./references/crdc-context.md
│   └─ Need more depth? → Load education-data-source-crdc skill
├─ schools/edfacts/* → Read ./references/edfacts-context.md
│   └─ Need more depth? → Load education-data-source-edfacts skill
├─ schools/meps/* → Load education-data-source-meps skill
├─ college-university/ipeds/* → Read ./references/ipeds-context.md
│   └─ Need more depth? → Load education-data-source-ipeds skill
├─ college-university/scorecard/* → Read ./references/scorecard-context.md
│   └─ Need more depth? → Load education-data-source-scorecard skill
├─ college-university/fsa/* → Load education-data-source-fsa skill
├─ college-university/nacubo/* → Load education-data-source-nacubo skill
├─ college-university/eada/* → Load education-data-source-eada skill
├─ college-university/pseo/* → Load education-data-source-pseo skill
├─ school-districts/saipe/* → Load education-data-source-saipe skill
└─ Multiple sources → Read ./references/data-relationships.md first

How do I interpret missing values?

What value do you see?
├─ In a CATEGORICAL column (grade, race, sex)?
│   └─ These use integer encoding, NOT coded missing values!
│       ├─ grade = -1 means Pre-K (NOT missing!)
│       ├─ race = 1-7 (NOT WH, BL, HI strings)
│       └─ sex = 1-2 (NOT M, F strings)
├─ In a NUMERIC column (enrollment, FTE, counts)?
│   ├─ -1 → Missing/not reported (treat as NULL)
│   ├─ -2 → Not applicable (exclude from that variable's analysis)
│   └─ -3 → Suppressed for privacy (cannot recover)
├─ null/blank?
│   └─ Source matters:
│       ├─ CCD, CRDC, EDFacts → Should use -1/-2/-3 codes
│       └─ Scorecard, MEPS, NACUBO → Use native nulls
├─ Ranges (e.g., "10-20") → EDFacts suppression bounds
└─ Unsure → Check source-specific reference file

What are the limitations?

What type of analysis are you doing?
├─ Cross-state comparison
│   ├─ K-12 assessments → INVALID (states not comparable)
│   ├─ K-12 other metrics → Check state reporting consistency
│   └─ College data → Generally valid (federal definitions)
├─ Time series
│   ├─ Check for definition changes
│   ├─ Check for ID changes (schools/districts merge/split)
│   └─ Check COVID-19 impact (2020-2021)
├─ Subgroup analysis
│   ├─ Check suppression rates
│   ├─ Smaller groups = more suppression
│   └─ Cannot impute suppressed values accurately
└─ Graduate outcomes
    ├─ IPEDS → First-time full-time only
    └─ Scorecard → Title IV recipients only

Universal Data Caveats

Portal Integer Encoding System

CRITICAL: The Education Data Portal uses integer codes, not string labels, for categorical variables. This applies to all sources.

Demographic Variable Encodings

Variable	Integer Values	NOT Strings
Race	1-7, 99 (total)	Not WH, BL, HI, AS, etc.
Sex	1 (Male), 2 (Female), 3 (Another gender, IPEDS 2022+), 4 (Unknown gender, IPEDS 2022+), 9 (Unknown), 99 (Total)	Not M, F
Grade	-1 to 13, 99 (total)	Not PK, KG, 01, etc.

Race codes:

Value	Meaning
1	White
2	Black
3	Hispanic
4	Asian
5	American Indian/Alaska Native
6	Native Hawaiian/Pacific Islander
7	Two or more races
8	Nonresident alien (postsecondary only)
9	Unknown
99	Total (all races)

Grade codes:

Value	Meaning
-1	Pre-K (SEMANTIC TRAP: NOT missing data!)
0	Kindergarten
1-12	Grades 1-12
13	Ungraded
99	Total (all grades)

SEMANTIC TRAP - Grade -1: In CCD enrollment data, grade = -1 means Pre-Kindergarten, NOT missing data. This is a common source of errors. Missing data in enrollment uses the separate coded value system (-1/-2/-3) only for numeric fields like enrollment counts, not for the grade categorical variable.

# WRONG - filters out Pre-K students!
df = df.filter(pl.col("grade") >= 0)

# RIGHT - Pre-K students have grade = -1
pre_k = df.filter(pl.col("grade") == -1)
k_12 = df.filter(pl.col("grade").is_between(0, 12))
total = df.filter(pl.col("grade") == 99)

Variable Names Are Lowercase

Portal variable names are lowercase, not the uppercase names from original NCES documentation:

enrollment not MEMBER or ENROLLMENT
grade not GRADE
fips not FIPS or STATE

Rate and Proportion Normalization

The Portal normalizes certain rate and proportion variables to a 0-1 scale, while the original IPEDS surveys report them as 0-100 percentages. This is a Portal transformation, not an IPEDS source issue.

Known affected variables:

Variable	Source Survey	Portal Scale	Original IPEDS Scale
`completion_rate_150pct`	GRS (Graduation Rates)	0-1	0-100
`retention_rate`	EF (Fall Enrollment / Retention)	0-1	0-100

Guidance:

Always check the actual range of rate variables after fetching -- if max <= 1.0, the variable is on a 0-1 scale and may need rescaling to 0-100 for interpretability
Do not assume all rate variables across all datasets are normalized -- this finding is specific to the IPEDS variables listed above
Quality checks testing value > 100 will not catch invalid data on 0-1 scaled variables; adjust thresholds accordingly (e.g., test value > 1.0 instead)

Missing Value Codes

Code	Meaning	How to Handle
-1	Missing/not reported	Treat as NULL; document missingness rate
-2	Not applicable	Exclude from analysis of that variable
-3	Suppressed (privacy)	Cannot be recovered; affects small-cell analyses
null/blank	Genuinely missing	Treat as NULL

IMPORTANT: Coded values (-1/-2/-3) apply to numeric measure columns (enrollment counts, FTE, etc.), NOT to categorical identifier columns like grade, race, or sex. Those use the integer encoding system above.

Missing Data Handling Varies by Source:

Source	Missing Data Pattern
CCD, CRDC, EDFacts	Use -1/-2/-3 coded values for numeric fields
Scorecard, MEPS, NACUBO	Use native `null` values
IPEDS	Mix of both (check specific variables)

Important: Filter coded values BEFORE calculating statistics:

# WRONG - includes coded values in mean
df["enrollment"].mean()

# RIGHT - exclude coded missing values
df.filter(pl.col("enrollment") >= 0)["enrollment"].mean()

Year Definitions

year refers to the FALL of the academic year
year=2020 means the 2020-21 school year
Graduation rates use cohort entry year (cohort started 4-6 years prior)
Finance data may use fiscal year (varies by institution)

Data Type	Year Interpretation
Fall enrollment	Fall of indicated year
Academic year totals	Full year starting fall of indicated year
Graduation rates	Cohort entry year (outcomes measured later)
Completions	Degrees awarded during indicated academic year

Suppression

Data is suppressed to protect student privacy:

Small cell sizes: Typically fewer than 5-10 students
Affects disaggregated data: Race, disability, gender breakdowns
More suppression in smaller schools: Rural areas most affected
Cannot be imputed accurately: Do not attempt to recover
Complementary suppression: Other cells may be suppressed to prevent calculation

State Reporting Variation

State education agencies interpret federal definitions differently:

Dropout definitions vary (CCD covers grades 7-12, CPS covers 10-12)
Average daily attendance calculated differently by state law
Discipline categories interpreted inconsistently
Missing data tends to cluster by state

Data Quality Checklist

Before analyzing any Education Data Portal data:

Check coded values: Filter out -1, -2, -3 before calculations
Understand year definition: Fall of academic year vs. cohort year
Note suppression rates: Calculate % suppressed by variable
Check definition changes: Compare codebooks across years
Verify identifier consistency: NCES IDs can change when schools/districts merge
Document state anomalies: Note any state-specific reporting issues
Check coverage: Not all schools appear in all sources
Consider COVID-19: 2020-2021 data may not be comparable to prior years

Quick Coverage Check

# Check missingness and suppression by state
df.group_by("fips").agg([
    pl.col("variable").filter(pl.col("variable") == -1).count().alias("missing"),
    pl.col("variable").filter(pl.col("variable") == -3).count().alias("suppressed"),
    pl.col("variable").count().alias("total")
])

Citation Requirements

Full Citation Format

Use for publications, reports, and formal documents:

[Dataset name(s)], Education Data Portal (Version X.X.X), 
Urban Institute, accessed [Month DD, YYYY], 
https://educationdata.urban.org/documentation/, 
made available under the ODC Attribution License.

Example:

Common Core of Data (CCD) School Directory, Education Data Portal 
(Version 0.20.0), Urban Institute, accessed January 15, 2026, 
https://educationdata.urban.org/documentation/, 
made available under the ODC Attribution License.

Short Citation Format

Use for visualizations, dashboards, and space-constrained contexts:

Source: [Dataset name(s)], Education Data Portal v.X.X.X, 
Urban Institute, ODC-By License.

Example:

Source: CCD School Directory, Education Data Portal v.0.20.0, 
Urban Institute, ODC-By License.

License Terms

License: Open Data Commons Attribution License (ODC-By) v1.0

Key requirements:

Must attribute the Urban Institute as data source
Must indicate if data was modified
May use for any purpose including commercial
May redistribute with attribution

Notification

Email [email protected] with any published work using the data. This helps the Urban Institute track usage and improve the portal.

Quick Reference: Source-Specific Caveats

Source	Key Limitation	Critical For	Quick Reference	Deep Dive
CCD	Public schools only; state reporting varies	K-12 enrollment, demographics	`./references/ccd-context.md`	`education-data-source-ccd`
IPEDS	First-time full-time students only for grad rates	College graduation analysis	`./references/ipeds-context.md`	`education-data-source-ipeds`
CRDC	Biennial; self-reported; underreporting	Equity/discipline analysis	`./references/crdc-context.md`	`education-data-source-crdc`
Scorecard	Title IV recipients only; earnings suppressed	Earnings/outcomes analysis	`./references/scorecard-context.md`	`education-data-source-scorecard`
EDFacts	State assessments NOT comparable across states	Achievement analysis	`./references/edfacts-context.md`	`education-data-source-edfacts`
SAIPE	Model-based estimates; no district CIs	District poverty	—	`education-data-source-saipe`
FSA	Federal aid only; timing varies	Student aid analysis	—	`education-data-source-fsa`
MEPS	Model estimates; 100% FPL only	School poverty (cross-state)	—	`education-data-source-meps`
NHGIS	Boundary changes over time	Geography linking	—	`education-data-source-nhgis`
EADA	Self-reported; NOT Title IX compliance	Athletics equity	—	`education-data-source-eada`
Campus Safety	Underreporting; comparability issues	Campus crime	—	`education-data-source-campus-safety`
PSEO	Experimental; partial state coverage	Employment outcomes	—	`education-data-source-pseo`

What Each Source Covers

Source	Universe	Update Frequency
CCD	All public schools and districts	Annual
IPEDS	All Title IV postsecondary institutions	Annual
CRDC	Sample/universe of public schools	Biennial
Scorecard	Title IV aid recipients	Annual
EDFacts	Public schools with state assessments	Annual

Data Lag Reference

Data availability lags behind the current year. As of January 2026:

Source	Survey Component	Typical Lag	Latest Available
IPEDS	Directory	~1 year	2023
IPEDS	Admissions-Enrollment	~2 years	2022
IPEDS	Fall Enrollment	~2-3 years	2021
IPEDS	Finance	~2-3 years	Varies
CCD	Directory/Enrollment	~1-2 years	2022
CCD	Finance	~2-3 years	2020
CRDC	All (biennial)	~1-2 years	2021
EDFacts	Assessments	~1-2 years	2020
EDFacts	Graduation Rates	~1-2 years	2020
SAIPE	Poverty estimates	~18 months	2023
Scorecard	Earnings/outcomes	~2-3 years	2020
MEPS	School poverty	~2-3 years	2019

Always verify year availability before building pipelines. Use mirror discovery endpoints (see mirrors.yaml) or filter downloaded data to confirm which years are present. See education-data-query skill for mirror-based fetch patterns.

Common Analysis Mistakes

DO NOT:

Compare state assessment scores across states (EDFacts)
- Each state has different tests and cut scores
Use IPEDS graduation rates to represent all students
- Only tracks first-time, full-time students
Assume Scorecard earnings represent all graduates
- Only covers Title IV aid recipients
Calculate statistics without filtering coded values
- -1, -2, -3 are not zeros; they corrupt calculations
Compare 2020-2021 data to prior years without noting COVID
- Testing waivers, discipline changes, enrollment shifts
Merge data across years assuming stable identifiers
- Schools and districts merge, split, and change IDs
Assume Portal rate variables are on a 0-100 percentage scale
- Some IPEDS rate variables (e.g., completion_rate_150pct, retention_rate) are normalized to 0-1 proportions in the Portal, even though the original IPEDS surveys use 0-100. Always check the actual range after fetching. See "Rate and Proportion Normalization" above.

DO:

Check suppression rates before disaggregating
Use within-state comparisons for assessment data
Document all data limitations in your analysis
Verify identifier stability for longitudinal analyses
Cite the data source properly

Cross-References

Variable definitions: Load education-data-explorer skill to understand what variables measure
Query assistance: Load education-data-query skill to re-fetch data with different parameters
Joining data: Read ./references/data-relationships.md for identifier mappings
Deep source context: Load the appropriate education-data-source-* skill for comprehensive methodology, historical changes, and detailed variable definitions
Source-specific gotchas: Load the relevant education-data-source-* skill for variable name mappings, data lags, and endpoint-specific behaviors

Topic Index

Topic	Location
Bureau of Indian Education schools	`./references/ccd-context.md`
Charter school coverage	`./references/ccd-context.md`
Chronic absenteeism	`./references/crdc-context.md`
Citation format	This file: Citation Requirements
COVID-19 data impact	`./references/crdc-context.md`
Discipline data	`./references/crdc-context.md`
Dropout definitions	`./references/ccd-context.md`
Earnings data limitations	`./references/scorecard-context.md`
Finance data (colleges)	`./references/ipeds-context.md`
GASB vs FASB accounting	`./references/ipeds-context.md`
Graduation rate caveats	`./references/ipeds-context.md`
Identifier relationships	`./references/data-relationships.md`
Joining tables	`./references/data-relationships.md`
LEAID format	`./references/data-relationships.md`
Locale codes	`./references/ccd-context.md`
Missing value codes	This file: Universal Data Caveats
NCESSCH format	`./references/data-relationships.md`
Net price calculation	`./references/ipeds-context.md`
ODC-By License	This file: Citation Requirements
OPEID vs UNITID	`./references/data-relationships.md`
Private schools	`./references/ccd-context.md` (not covered)
Proficiency data	`./references/edfacts-context.md`
Race category changes	`./references/ccd-context.md`
Sampling (CRDC)	`./references/crdc-context.md`
State assessment comparability	`./references/edfacts-context.md`
State FIPS codes	`./references/data-relationships.md`
Student financial aid	`./references/ipeds-context.md`
Suppression	This file: Universal Data Caveats
Title IV institutions	`./references/ipeds-context.md`
Transfer students	`./references/ipeds-context.md`
UNITID changes	`./references/ipeds-context.md`
Year definitions	This file: Universal Data Caveats

When it activates

Automatically, when a task matches. Use when interpreting Portal data before analysis. Routes to source-specific skills.

About this skill

Automatic

Data Analyst 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

brycewang-stanford / Auto-Empirical-Research-Skills

skills/17-DAAF-Contribution-Community-daaf/dot-claude/skills/education-data-context/SKILL.md

View on GitHub ↗

License NOASSERTIONImported 8/2/2026Version v1Bundled 6 files

Illustrative Want to see what brycewang-stanford/education-data-context 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

Education Data Context

Why Data Context Matters

Data Provenance: The Education Data Portal

Reference File Structure

Quick Context (This Skill)

Deep-Dive Source Skills (Comprehensive Documentation)

Decision Trees

What data source did I pull from?

How do I interpret missing values?

What are the limitations?

Universal Data Caveats

Portal Integer Encoding System

Demographic Variable Encodings

Variable Names Are Lowercase

Rate and Proportion Normalization

Missing Value Codes

Year Definitions

Suppression

State Reporting Variation

Data Quality Checklist

Quick Coverage Check

Citation Requirements

Full Citation Format

Short Citation Format

License Terms

Notification

Quick Reference: Source-Specific Caveats

What Each Source Covers

Data Lag Reference

Common Analysis Mistakes

DO NOT:

DO:

Cross-References

Topic Index

When it activates

SkillSafety

Source

Related skills

More in the registry