[codex] Add evidence-backed native Python idiom advice

[GTrunSec]

Summary

• add an evidence note that ties official Python guidance, mature GitHub project practice, and harness policy scope together
• extend parser-owned function control-flow facts with simple manual collection and predicate loop signals
• add advisory PY-AGENT-R011 so agents prefer comprehensions, generator expressions, built-ins, or named iterator pipeline helpers when public functions spell boilerplate loops

Policy rationale

This keeps ruff/mypy/pytest as the normal tool substrate and uses the harness for the gap those tools do not own: compact reasoning-tree advice for LLM code repair. The new rule is parser-first and advisory; explicit loops remain valid for effects, complex state, debugging, or measured performance work.

Evidence sources

• PEP 20 / PEP 8
• Python docs for match/case, comprehensions, itertools, built-ins, dataclasses, and typing
• GitHub source/config review of pydantic, pytest, httpx, and black

Validation

• uv run --group test ruff format --check src tests docs
• uv run --group test ruff check src tests
• uv run --group test pytest tests/unit/python_lang_parser/test_control_flow.py tests/unit/harness/test_agent_algorithm_policy.py tests/unit/harness/test_policy_contract.py tests/unit/harness/test_agent_policy.py tests/unit/harness/test_parser_boundary_contract.py -q
• uv run --group test pytest tests -q
• uv run --group test python-project-harness .
• uv run --group test python-project-harness --agent-snapshot .
• uv build
• git diff --check

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6455254275

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Invalidate accumulator bindings after name reassignment

This pass keeps collection_bindings forever once a name is initialized to []/{}/set(), so later loops can be flagged even after that name is reassigned to a non-empty or unrelated value. For example, acc = []; acc = preload(); for x in xs: acc.append(x) is reported as a manual accumulator loop even though it is no longer a simple empty-accumulator pattern. That produces incorrect PY-AGENT-R011 advice and undermines the rule’s stated conservative behavior.

Useful? React with 👍 / 👎.