Ready to supercharge your AI agents?

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

hipDNN Superbuild Test Runner · rocm | DecimalAI

Testingby rocm

hipDNN Superbuild Test Runner

Run tests against an existing hipDNN superbuild. Supports per-component selection (hipdnn, miopen-provider, hipblaslt-provider, hip-kernel-provider, integration-tests), unit/integration/external-integration scope, and gtest filtering. Reproduces the cross-provider external-integration-check suite. Handles Windows DLL PATH automatically.

Never made a case worseMaintained

SkillScore

SkillScore86/100

Test results 86

Safety

3/3 safety checks passed

Code scan
Intent
Content

What it changes

29%86%of test cases passed +57 pts

Without skill6 of 21

With skill18 of 21

12 fixed6 passed either way3 still failing0 broke1 errored

Model 11/4/2026v722 attempted

Track record

Agent runs: 0; All time

See one task, with and without the skill

One test taskI don't have a build directory yet for hipdnn. Please configure the CMake project with release settings and build the entire superbuild from source.

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 rocm-hipdnn-superbuild-test
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("rocm-hipdnn-superbuild-test")
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 rocm-hipdnn-superbuild-test

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/rocm-hipdnn-superbuild-test/SKILL.md https://app.decimal.ai/s/rocm-hipdnn-superbuild-test/SKILL.md

Copy SKILL.md

Save it in your project, as

.claude/skills/rocm-hipdnn-superbuild-test/SKILL.md

This gets SKILL.md only. The skill's 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from agents import Agent, Runner

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from openai import OpenAI

client = OpenAI()
skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
import anthropic

client = anthropic.Anthropic()
skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/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/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from langchain.agents import create_agent

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

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

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from pydantic_ai import Agent

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from deepagents import create_deep_agent

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

import asyncio
from pathlib import Path
from agent_framework import Agent

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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("rocm-hipdnn-superbuild-test")
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/rocm-hipdnn-superbuild-test/SKILL.md

Give it to your agent

from pathlib import Path
from strands import Agent

skill = Path(".agents/skills/rocm-hipdnn-superbuild-test/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 4 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 rocm-hipdnn-superbuild-test

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/rocm-hipdnn-superbuild-test/SKILL.md https://app.decimal.ai/s/rocm-hipdnn-superbuild-test/SKILL.md

Copy SKILL.md

Save it in your project, as

.agents/skills/rocm-hipdnn-superbuild-test/SKILL.md

This gets SKILL.md only. The skill's 4 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/rocm-hipdnn-superbuild-test/SKILL.md

Read the summary as JSON

curl -fsSL https://app.decimal.ai/s/rocm-hipdnn-superbuild-test.json

This gets SKILL.md only. The skill's 4 other files are not included.

Other ways to add it

Other setups docs

You don't need a DecimalAI account.

Version: v7, tested 11/4/2026
Source: rocm/rocm-libraries ↗
License: Not recorded

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

hipDNN Superbuild Test Runner

Use this skill when the user asks to test an existing hipDNN superbuild. It does not configure or build the project. If no superbuild exists, tell the user to build first with hipdnn-superbuild.

Inputs

Infer options from the user request:

Component: hipdnn, miopen, hipblaslt, hip-kernel, integration-tests, or all; default all
Scope: unit, integration, external-integration, or all; default unit. external-integration covers the cross-provider hipdnn_integration_tests suite (the <provider>-external-integration-check targets)
Filter: optional gtest filter; when present, run test binaries directly
Verbose: use verbose test targets when requested
Keep going: continue after failures only when requested
ROCm path: optional ROCM_PATH=<path> override; Linux defaults to /opt/rocm
Jobs: optional explicit parallelism only when the user requests it and active workspace instructions permit it

Workflow

Determine the repository root:
```
git rev-parse --show-toplevel
```
Resolve paths:
- Build directory: honor active workspace instructions first; otherwise use <repo-root>/build.
- Binary directory: <build-dir>/bin. <binary-path> below is the full path to a test executable under it.
- <installed-ctest-root>: the CTest root of an installed tree, which for this provider is <install-prefix>/bin/hip_kernel_provider, not the prefix itself.
- <build-ctest-root>: <build-dir>/dnn-providers/hip-kernel-provider/src, the build-tree registration of the categorized engine-pinned suites, with absolute paths. List one engine's with ctest --test-dir <build-ctest-root> -N -L <engine-label>, where the label is the engine ID with every character outside [A-Za-z0-9_.-] mapped to _ (for example hipkernel_Gfx950AttentionDense). The provider's install_CTestTestfile.cmake (or CTestTestfile.cmake.install) in the build tree is install-only: CTest does not read it there. The per-arch census shard runs in either tree, at <build-dir>/lib/hipdnn_plugins/engines/test_arch_content/hip-kernel-provider/census/<arch> in the build tree (<build-dir>/bin/... on Windows) and the same relative path under the install prefix.
- <PY>: for the corpus-sweep command below, the Python interpreter the active workspace or repository instructions mandate, otherwise the active venv's python. Resolve it once and substitute the full path.
- <GEN>: the IngestorGenerator root, <repo-root>/projects/hipdnn/tools/IngestorGenerator.
- Helper scripts: skills are host-level, not tied to a repo checkout — default to the scripts bundled with the skill you were invoked from (<skill-directory>/scripts), even when working inside a repo or worktree. Do NOT run the <repo-root>/projects/hipdnn/tools/ai/skills/hipdnn-superbuild-test/scripts copy just because a checkout is present: it can be a stale stub (on develop) or an unmerged in-progress version (on a feature branch). Use the source-checkout copy only when actively developing this skill itself to exercise your in-progress edits, when the invoked skill has no bundled scripts/ directory, or for ingestor engine create/extend work, where the helper revision must match the checkout being built.
Verify the superbuild exists:
```
ls <build-dir>/build.ninja
```
Stop if it is missing.
Resolve ROCm path on Windows:
```
python3 <scripts>/windows_rocm_setup.py --repo-root <repo-root> [--rocm-path <path>]
```
Parse ROCM_PATH=... from stdout and set ROCM_BIN=<rocm-path>/bin. Skip this step on Linux unless the user supplied an override. On Windows, always pass the resolved ROCM_BIN to cmake_run.py (steps 6-8) via --rocm-bin: it is required both for the runtime PATH and for staging the wheel's System32-shadowed DLLs app-local (see Notes).
Discover CMake test targets:
```
python3 <scripts>/discover_test_targets.py --build-dir <build-dir> --component <component> --scope <scope>
```
The helper prints <component>:<target> lines. It also handles the hip-kernel-provider path-qualified target naming. With --scope external-integration (or all) it emits one <component>:<target> line and one <component>:command:<cmdline> line for each registered external suite. A provider can register several suites. For example, hip-kernel-provider registers one suite for each engine: hip-kernel-provider-external-integration-check for HIP_MLOPS_ENGINE and hip-kernel-provider-asm-sdpa-external-integration-check for ASM_SDPA_ENGINE. Do not assume a single line. Each command is the resolved cross-provider hipdnn_integration_tests invocation (with --test-article/--test-engine/--test-config) read from the generated CTestTestfile.cmake, with any baked-in --gtest_filter removed so that you can supply your own. The --test-engine value shows which suite a line belongs to. If the helper reports that Ninja target discovery failed, treat that as an invalid or stale build directory and stop with the helper's diagnostic. If discovery succeeds but no targets match, report that the requested component or scope is not present in the existing superbuild.

For an ingestor engine the discovery component is hip-kernel, not hip-kernel-provider. A helper's first provider-prefixed command need not be the requested engine's registration; inspect the actual installed CTest entry before executing it. Replace <your-bundle-ctest-target> with the name your own registration creates. The gfx942 dense names here are illustrative, not targets to copy: unless your build registers hip_kernel_provider_gfx942_attention_dense_gpu_ref_integration_tests, copying it verbatim fails the second command under --no-tests=error:
```
ctest --test-dir <installed-ctest-root> -N -V \
  -R '^<your-bundle-ctest-target>$'
ctest --test-dir <installed-ctest-root> --no-tests=error -V \
  -R '^<your-bundle-ctest-target>$'
```
Require your bundle's exact registration, its engine pin (hipkernel:Gfx942AttentionDense in the illustration), current installed executable/plugin/config paths and intended quick/standard cases. Missing registration, wrong pin, zero selected cases, all-skipped support or failed numerical comparisons fail this gate. A broad component PASS is not exact-engine evidence.
Run tests through cmake_run.py when no gtest filter is requested:
```
python3 <scripts>/cmake_run.py --build-dir <build-dir> --target <target> [--rocm-path <path>] [--rocm-bin <path>] > <log> 2>&1
```
Add --jobs <N> only when explicit jobs are both requested and permitted. For verbose mode, append -verbose to the target name.
Run direct binaries when a gtest filter is requested:
```
python3 <scripts>/cmake_run.py --build-dir <build-dir> --binary <binary-path> --gtest-filter "<filter>" [--extra-arg=<flag> ...] [-- <passthrough args>] [--rocm-path <path>] [--rocm-bin <path>] > <log> 2>&1
```
Use the component-to-binary mapping below to choose binaries. cmake_run.py accepts arbitrary passthrough flags for the binary: simple values via repeatable --extra-arg (use --extra-arg=--flag for flag-like values), or an entire flag list after a literal --. Passing multiple tokens inside --binary is rejected with a clear error.
Reproduce the cross-provider external-integration suite (--scope external-integration):
- To run the whole suite exactly as CI does, build the custom target:
```
python3 <scripts>/cmake_run.py --build-dir <build-dir> --target <provider>-external-integration-check [--rocm-path <path>] [--rocm-bin <path>] > <log> 2>&1
```
- To run with a custom gtest filter, take the <component>:command:<cmdline> line from step 5, run the first token as --binary and the rest after --, adding your own --gtest-filter:
```
python3 <scripts>/cmake_run.py --build-dir <build-dir> --binary <hipdnn_integration_tests> -- <--test-article ... --test-engine ... --test-config ...> --gtest_filter=<filter> > <log> 2>&1
```
- Before reporting the result, load the hipdnn-integration-testing skill. This suite is the only place where exit code 0 is often meaningless. That skill applies the suite's own documentation to the output. This includes the signals that the suite prints but that never fail a run.
- An ingestor engine's explicit-selection C++ cases run inside the shared hip_kernel_provider_integration_tests binary (CTest entry hip_kernel_provider_integration_tests_quick_suite, which runs every engine's cases); the ingestor RUNBOOK stage 5 gives the filtered command.
For every command, keep full output in a log and show only a short tail on failure. Track pass/fail per component. Stop at the first failure unless keep-going was requested. For external-integration runs, report the Passed:/Skipped:/Failed: counts from the binary's "TEST COVERAGE SUMMARY". Never report only the exit code. A run that skips 100% of tests is green, but it does not show that the engine still works.

Direct Binary Mapping

Component	Unit Binaries	Integration Binaries	External Integration (cross-provider)
`hipdnn`	`hipdnn_backend_tests`, `hipdnn_frontend_tests`, `hipdnn_data_sdk_tests`, `hipdnn_flatbuffers_sdk_tests`, `hipdnn_plugin_sdk_tests`, `hipdnn_test_sdk_tests`	`hipdnn_public_backend_tests`, `hipdnn_public_frontend_tests`, `hipdnn_backend_logging_shutdown_tests`	—
`miopen`	`miopen_plugin_tests`	`miopen_plugin_integration_tests`	`miopen-provider-external-integration-check` (`hipdnn_integration_tests` + `miopen_plugin`, engine `MIOPEN_ENGINE`)
`hipblaslt`	`hipblaslt_plugin_tests`	`hipblaslt_plugin_integration_tests`	`hipblaslt-provider-external-integration-check` (`hipdnn_integration_tests` + `hipblaslt_plugin`, engine `HIPBLASLT_ENGINE`)
`hip-kernel`	`hip_kernel_provider_tests`	`hip_kernel_provider_integration_tests`	`hip-kernel-provider-external-integration-check` when present
`integration-tests`	`hipdnn_integration_tests_unit_tests`	`hipdnn_integration_tests`, `hipdnn_gpu_ref_tests`	—

The exact article/engine/config for the external suite is resolved at build time. Use discovery for available commands, then inspect the specific CTest registration when proving a named engine; a generic command: line is not that proof.

Packaging probes

When the build was configured with HIPKERNELPROVIDER_ENABLE_PACKAGING_PROBES=ON (superbuild-only, OFF by default), the probe ctest entries are hkp-probe-<arch>, one per architecture explicitly named under the production descriptor root that GPU_TARGETS does not include (automatic; none when the build targets them all), hkp-probe-<NAME>, one per named hkp_add_packaging_probe() line in probes.cmake, and hkp-probe-tools. A full ctest run includes them; to run them alone, select with the anchored -R '^hkp-probe-' and --no-tests=error. -DHIPKERNELPROVIDER_MIN_PACKAGING_PROBES=N fails configure when fewer than N probes are declared (default 0).

In the superbuild CI lanes the probes run once, inside the "Run tests" ctest, with N=1. A pass does not cover device behavior, install staging, UKDs the probe does not pack, or standalone-UKD references.

Ingestor proof boundaries

The ingestor RUNBOOK is the sole ordered create/extend workflow and states how each gate is invoked. What matters here is what a pass does not establish: device_probe.py success, in either mode, is not dispatch.

Native host proof executes actual typed provider registrations and descriptor loading, then checks the finalized emitted inventory. Use a fresh process, explicit HIPDNN_TEST_EXPECTED_ARCH from configured packaging architectures, the corresponding shard and a nonempty exact host-test selection. Missing or unknown architecture selection, wrong-arch data, absent/extra identities and wrong runtime source kind fail; packaged runtime source kind is KPACK. Source-text symbol matching and structural validation cannot certify native hooks, and host loading cannot prove device dispatch.

Numerical acceptance needs a capable independent reference for the actual graph, and neither SDPA reference supports a sink UID. Record BLOCKED when no capable reference exists; a skip, automatic fallback exhaustion or unverified golden output cannot pass.

The corpus sweep interface is <PY> <GEN>/tools/sweep.py --config <absolute-YAML> with configs/reduction.sweep.yaml.example. Correctness is separate from timing, engine attribution is exact, and resume is bound to current input content. SWEEP_DONE is validated completion; explicit correctness.enabled: false yields SWEEP_TIMING_ONLY, never final acceptance; unmet gates yield SWEEP_INCOMPLETE. After tuning or regeneration, repeat artifact/native/device/corpus gates against the final installation and complete the per-corpus runtime outcome join. Passing only unchanged baseline cases cannot establish that an extension's new variant served.

Report

Summarize per-component results:

hipdnn:
  hipdnn-unit-check: PASS
miopen-provider:
  miopen-provider-unit-check: FAIL (see <log>)

If a requested component has no matching target, say that it was not present in the existing superbuild and name the preset or component likely needed.

Notes

scripts/cmake_run.py, scripts/discover_test_targets.py, scripts/windows_rocm_setup.py, and scripts/stage_shadowed_dlls.py are bundled in this skill so linked and copied installs work independently.
Windows DLL loading is handled by cmake_run.py, which sets PATH in Python's subprocess environment before launching CMake or test binaries.
Windows shadowed-DLL staging: before launching any target or binary on Windows, cmake_run.py stages the wheel's amd_comgr.dll and HIP runtime amdhip64_<N>.dll into <build-dir>/bin (via stage_shadowed_dlls.py) so the process does not load the driver's stale System32 copies. This happens on every Windows run, not just for a specific kernel path. Stale comgr breaks MIOpen's runtime JIT (GCN-assembly Winograd solvers are the common failure: [BuildAsm] comgr status = ERROR / unknown emulation: no-xnack); a stale HIP runtime makes the wheel's rocBLAS fault with SEH exception with code 0xc0000005 in MIOpen's GEMM conv solvers (e.g. GemmFwd1x1_0_1), after which the run can hang. This needs --rocm-bin to be passed. A copy is skipped when the staged DLL already matches the wheel's (PE version, else content hash), so it adds no cost on repeat runs. Disable with --no-stage-shadowed-dlls if ever needed. To confirm which comgr loaded, run a test with MIOPEN_LOG_LEVEL=7 MIOPEN_ENABLE_LOGGING=1 and grep for COMgr v. (a low version indicates the stale System32 copy; the wheel's is newer). The driver's System32 HIP runtime announces itself with a HIP Library Path: C:\windows\SYSTEM32\amdhip64_<N>.dll line at startup; the wheel's runtime prints no such line. MIOpen's AmdRocmMetadataVersionDetect HIP version is the build-time version and does not tell you which runtime loaded.
The build's stage_shadowed_rocm_dlls target (projects/hipdnn/cmake/WindowsDllStaging.cmake and dnn-providers/cmake/WindowsDllStaging.cmake) is the primary mechanism for app-local staging. cmake_run.py's staging is kept on purpose rather than as a leftover: it covers build trees configured before that target existed, and a newly discovered System32-shadowed DLL can be added to stage_shadowed_dlls.py right away, ahead of the matching CMake change. When you add a DLL to one, add it to the other.
Integration tests require an AMD GPU. Unit scope is the default for CPU-only validation.
Prefer running test binaries through cmake_run.py (it wires PATH/ROCM_PATH for the loader); pass extra binary flags via --extra-arg/-- <args> rather than folding them into --binary.

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?

hipDNN Superbuild Test Runner

SkillScore

Safety

What it changes

Track record

Use this skill

hipDNN Superbuild Test Runner

Inputs

Workflow

Direct Binary Mapping

Packaging probes

Ingestor proof boundaries

Report

Notes

See also

SkillSafety

Source

Related skills

Ready to supercharge your AI agents?

SkillScore

Safety

What it changes

Track record

Use this skill

hipDNN Superbuild Test Runner

Inputs

Workflow

Direct Binary Mapping

Packaging probes

Ingestor proof boundaries

Report

Notes

See also

SkillSafety

Source

Related skills

More in the registry