Greptile Summary

This PR adds normalize_output_format() to skills/openai-image-gen/scripts/gen.py to validate and normalize the --output-format CLI argument for gpt-image models. The change is well-scoped, logically correct, and fits naturally into the existing pre-flight validation pattern.

Key strengths:

• jpg is correctly aliased to jpeg as a user convenience.
• Unsupported formats raise a ValueError caught in main() and returned as exit code 2, consistent with other early-exit error handling.
• The normalized value is applied consistently to both the API request payload and the output filename extension.

Minor test gaps:

• Valid format pass-through ("png" → "png", etc.), case normalization ("PNG" → "png"), and empty-string cases for gpt-image models lack test coverage. These should be added to prevent regressions.

Confidence Score: 4/5

• Safe to merge. The core validation logic is correct with no regressions; minor test gaps are non-blocking.
• The implementation is sound and integrates cleanly with existing error handling. Valid-format pass-through tests are missing, which represents a coverage gap but not a correctness issue. The critical paths (invalid-value rejection, jpg→jpeg aliasing, non-gpt-image passthrough) are tested.
• skills/openai-image-gen/scripts/test_gen.py - add tests for valid format pass-through scenarios

_{Last reviewed commit: 4189b01}

Pushed a follow-up cleanup to this branch.

The diff now includes the added pass-through / empty-input coverage, warning coverage for incompatible models, a changelog entry with attribution, and the Python import-order fix that was tripping skills-python. I reran python3 -m ruff check skills/openai-image-gen/scripts/test_gen.py and python3 -m pytest -q skills/openai-image-gen/scripts/test_gen.py locally before pushing.

Rebased this branch onto current main after #36762 merged.

The overlap in gen.py, test_gen.py, and the changelog is resolved on the new head. The validation path now uses one shared helper for --background, --style, and --output-format, keeps the explicit warning behavior for incompatible models, preserves the changelog attribution, and keeps the targeted test coverage on the rebased diff.

Local verification on the rebased branch:

• python3 -m ruff check skills/openai-image-gen/scripts/gen.py skills/openai-image-gen/scripts/test_gen.py
• python3 -m pytest -q skills/openai-image-gen/scripts/test_gen.py (22 passed)