agentleFS
MicrosoftKnown publishermicrosoft/amplifier-bundle-skills/modules/tool-skills/tests/fixtures/skills/python-standards/SKILL.md

python-standards

Skill12 starsChanged 7 months agoMITView on GitHub

Python coding standards for Amplifier including type hints, async patterns, error handling, and formatting. Use when writing Python code for Amplifier modules.

We split this skill into the rules it gives an agent

5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Files must end with newline: - Add blank line at EOF

    line 117
  2. Very specific:

    Never manually edit pyproject.toml dependencies: - Use uv add

    line 139
  3. Very specific:

    Run make check before committing

    line 209
  4. Distinctive:

    Keep modules focused and simple

    line 210
  5. Distinctive:

    Test at the protocol level

    line 212

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Python Coding Standards

Type Hints

ALL functions must have complete type hints:

from typing import Any

async def process_data(items: list[str], config: dict[str, Any]) -> dict[str, Any]:
    """Process data items with configuration."""
    results = {}
    for item in items:
        results[item] = await transform(item, config)
    return results

Include type hints for self:

class MyClass:
    def __init__(self: "MyClass", name: str) -> None:
        self.name = name

    async def process(self: "MyClass") -> str:
        return f"Processing {self.name}"

Async Patterns

All I/O operations must be async:

# Good
async def read_file(path: Path) -> str:
    content = path.read_text()  # For now, sync is OK
    return content

# Better (when using async libraries)
async def read_file(path: Path) -> str:
    async with aiofiles.open(path) as f:
        return await f.read()

Use asyncio.gather for parallel operations:

async def process_files(files: list[Path]) -> list[dict]:
    tasks = [process_file(f) for f in files]
    return await asyncio.gather(*tasks)

Error Handling

Return errors, don't raise:

from amplifier_core import ToolResult

async def execute(self: "MyTool", input: dict[str, Any]) -> ToolResult:
    """Execute tool operation."""
    try:
        result = await self._process(input)
        return ToolResult(success=True, output=result)
    except ValueError as e:
        logger.error(f"Validation error: {e}")
        return ToolResult(success=False, error={"message": str(e)})

Provide clear error messages:

# Good
return ToolResult(
    success=False,
    error={"message": f"File not found: {path}"}
)

# Bad
return ToolResult(
    success=False,
    error={"message": "Error"}
)

Formatting

Line length: 120 characters

Import organization:

# Standard library
import asyncio
import logging
from pathlib import Path
from typing import Any

# Third-party
import yaml
from pydantic import BaseModel

# Local/Amplifier
from amplifier_core import ModuleCoordinator, ToolResult

Files must end with newline - Add blank line at EOF

Use ruff for formatting:

uv run ruff format .
uv run ruff check . --fix

Dependencies

Use uv for dependency management:

# Add dependency
cd amplifier-module-tool-mytool
uv add package-name

# Add dev dependency
uv add --dev pytest ruff pyright

Never manually edit pyproject.toml dependencies - Use uv add

Testing

Test behavior at protocol level:

import pytest
from amplifier_core.testing import TestCoordinator

@pytest.mark.asyncio
async def test_tool_basic():
    """Test basic tool functionality."""
    coordinator = TestCoordinator()

    # Mount module
    await mount(coordinator, {"timeout": 10})

    # Get and test
    tool = coordinator.get("tools", "my-tool")
    result = await tool.execute({"param": "value"})

    assert result.success
    assert "expected" in result.output

Test pyramid: 60% unit, 30% integration, 10% end-to-end

Common Pitfalls

Don't use blocking I/O

# Bad
content = requests.get(url).text

# Good
async with aiohttp.ClientSession() as session:
    async with session.get(url) as response:
        content = await response.text()

Don't modify module internals from outside

# Bad
tool._internal_state = new_value

# Good
await tool.execute({"operation": "update", "value": new_value})

Don't put logic in mount()

# Bad
async def mount(coordinator, config):
    tool = MyTool()
    await tool.initialize_database()  # Heavy logic
    await coordinator.mount("tools", tool)

# Good
async def mount(coordinator, config):
    tool = MyTool(config)  # Light initialization only
    await coordinator.mount("tools", tool)
    # Heavy logic happens in execute(), not mount()

Remember

  • Run make check before committing
  • Keep modules focused and simple
  • Document public interfaces
  • Test at the protocol level

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.