Agent Skills: Snake Polish — Implementation Phase

<file_paths>$ARGUMENTS</file_paths>

Snake Polish — Implementation Phase

Execute the implementation plan from /python-engineering:stinkysnake. Implement functions following the modernization plan, run tests iteratively until all pass.

Arguments

<file_paths/>

Prerequisites

Before invoking this skill, ensure:

1. /python-engineering:stinkysnake phases 1-8 completed
2. Modernization plan reviewed and refined
3. Interfaces designed and documented
4. Failing tests written by python-engineering:python-pytest-architect

Instructions

Step 1: Load Context

Read the stinkysnake plan artifacts:

ARTIFACTS TO LOAD:
- [ ] Modernization plan (Phase 3 output)
- [ ] Plan review feedback (Phase 4-5 output)
- [ ] Interface definitions (Phase 7 output)
- [ ] Failing test files (Phase 8 output)

Step 2: Verify Test Baseline

Run tests to confirm failing state:

uv run pytest <file_paths/> -v --tb=short 2>&1 | head -100

Expected: Tests fail because implementations don't exist yet.

If tests pass: Stop. Implementation already complete or tests are not testing the right things.

Step 3: Implementation Order

Follow this implementation sequence:

IMPLEMENTATION ORDER:
1. Type definitions (TypeAlias, TypedDict, Protocol)
2. Data structures (dataclass, Pydantic models)
3. Utility functions (pure functions, no side effects)
4. Core business logic
5. Integration points (API clients, file I/O)
6. Entry points (CLI commands, handlers)

Step 4: Implement Following Plan

For each planned change:

FOR EACH IMPLEMENTATION ITEM:
1. Read the interface/protocol definition
2. Read the failing test(s) for this component
3. Implement the function/class
4. Run targeted tests: uv run pytest -k "test_name" -v
5. If fails: debug and fix
6. If passes: move to next item

Step 5: Modern Python Patterns

Apply these patterns during implementation:

Type Annotations

# Use modern union syntax
def process(data: str | None) -> dict[str, Any]:
    ...

# Use TypeGuard for narrowing
def is_valid_user(obj: object) -> TypeGuard[User]:
    return isinstance(obj, dict) and "id" in obj

Protocol-Based Design

from typing import Protocol

class Serializable(Protocol):
    def to_dict(self) -> dict[str, Any]: ...

def save(item: Serializable) -> None:
    data = item.to_dict()
    ...

Dataclass Patterns

from dataclasses import dataclass, field

@dataclass(slots=True, frozen=True)
class Config:
    name: str
    options: list[str] = field(default_factory=list)

Pydantic for Validation

from pydantic import BaseModel, Field

class APIResponse(BaseModel):
    status: int = Field(ge=100, le=599)
    data: dict[str, Any]

    model_config = {"strict": True}

Modern Libraries

# httpx for async HTTP
async with httpx.AsyncClient() as client:
    response = await client.get(url)

# orjson for fast JSON
data = orjson.loads(response.content)
output = orjson.dumps(result, option=orjson.OPT_INDENT_2)

# tomlkit for TOML with comments preserved
doc = tomlkit.parse(content)
doc["section"]["key"] = value

Step 6: Iterative Test Loop

After each implementation batch:

# Run full test suite
uv run pytest <file_paths/> -v --tb=short

# If failures remain, focus on failing tests
uv run pytest <file_paths/> -v --tb=long -x  # Stop on first failure

Step 7: Static Analysis Verification

Before completion, verify code quality:

uv run prek run --files $ARGUMENTS
# Fallback when no .pre-commit-config.yaml:
# uv run ruff check $ARGUMENTS

Fix any issues that arise.

Step 8: Final Test Run

Confirm all tests pass:

uv run pytest <file_paths/> -v --cov --cov-report=term-missing

Success Criteria:

• All tests pass
• No type errors
• No lint errors
• Coverage meets project threshold (typically 80%+)

Completion

When all tests pass and static analysis is clean:

1. Report implementation summary
2. List any deferred items or technical debt
3. Reference documentation updates needed (from Phase 6)

Error Handling

Test Failures That Indicate Test Bugs

If a test failure appears to be a test bug rather than implementation bug:

1. Document the suspected test issue
2. Check the test against the interface specification
3. If test is wrong: fix the test, document the fix
4. If unclear: flag for review, continue with other implementations

Blocked Implementations

If an implementation is blocked:

1. Document the blocker
2. Check if it's a dependency ordering issue
3. If external dependency: note and continue with independent items
4. If architectural issue: flag for plan revision

References

• ../stinkysnake/SKILL.md - Parent workflow
• ../../agents/python-cli-architect.md - Implementation agent
• ../python3-core/SKILL.md - Modern Python patterns