tinkermonkey/rounds-check

Type Check & Lint

Quick-reference skill for rounds - runs mypy type checking and ruff linting to ensure code quality.

Usage

/rounds-check

Purpose

Validates code quality for the rounds continuous error diagnosis system by:

1. Type checking with mypy - Ensures 100% type annotation compliance (required by CLAUDE.md:45)
2. Linting with ruff - Enforces Python code style and catches common errors

This skill is critical because rounds uses strict type safety as a core architectural principle:

• All functions must be type-annotated with Python 3.11+ syntax
• Domain models use @dataclass(frozen=True) for immutability
• All I/O uses async def with proper async/await patterns
• Ports define abstract interfaces with precise type signatures

Implementation

Runs the following commands from the project's development dependencies:

1. Type Check with mypy

mypy rounds/

What it checks:

• All functions have type annotations (rounds/core/models.py:13-40)
• Async/await usage is correct (rounds/core/ports.py:15-85)
• Port interfaces match adapter implementations
• No Any types in domain layer (rounds/core/)
• Frozen dataclasses used correctly (rounds/core/models.py)

2. Lint with ruff

ruff check rounds/

What it checks:

• Import order (standard lib → third-party → local, per CLAUDE.md:125)
• Snake_case for files/functions, PascalCase for classes (CLAUDE.md:111-117)
• Unused imports and variables
• Line length and formatting
• F-string usage and comprehensions
• Error handling patterns

Examples

Run both checks

/rounds-check

Expected output (clean run):

✓ Running type check with mypy
Success: no issues found in 15 source files

✓ Running lint check with ruff
All checks passed!

Example failures and fixes

mypy error - missing type annotation:

rounds/core/fingerprint.py:42: error: Function is missing a return type annotation

Fix: Add return type to function signature

ruff error - import order:

rounds/adapters/store/sqlite.py:5:1: I001 Import block is un-sorted or un-formatted

Fix: Reorder imports (standard lib → third-party → local)

Key Files Checked

Based on project architecture:

Core domain layer (rounds/core/):

• models.py - Immutable domain entities (Signature, Diagnosis, ErrorEvent)
• ports.py - Abstract port interfaces (8 ports total)
• fingerprint.py - Error fingerprinting service
• triage.py - Error classification service
• investigator.py - Investigation orchestration
• poll_service.py - Polling loop service
• management_service.py - CLI/webhook operations

Adapter layer (rounds/adapters/):

• telemetry/signoz.py, jaeger.py, grafana_stack.py
• store/sqlite.py
• diagnosis/claude_code.py
• notification/stdout.py, markdown.py, github_issues.py
• scheduler/daemon.py
• webhook/http_server.py, receiver.py
• cli/commands.py

Composition root:

• main.py - Dependency injection and entry point
• config.py - Pydantic settings with environment variables

Why This Matters

From CLAUDE.md:35-42:

All code must be type-annotated with Python 3.11+ syntax Use from typing import ... for complex types Use TypeAlias for custom type definitions Frozen dataclasses for immutable domain objects

Type safety and code quality are non-negotiable in rounds because:

• Hexagonal architecture requires precise port/adapter contracts
• Async I/O patterns must be verified (no accidental blocking calls)
• Immutable domain models prevent state corruption
• LLM diagnosis costs real money - bugs are expensive

---

This skill was automatically generated from rounds project conventions.