deephaven-mcp: Skill for Claude Code

Review like a senior engineer. Every finding must answer three questions in concrete terms: what is wrong, what is better, why the change is worth its cost. All topics are fair game — correctness, security, design, duplication, clarity, naming, structure, tests. Be conscious of suggestions that do not serve a purpose.

Perform a comprehensive review of the specified Python file as it currently exists.

1. Design: Is the code well-structured and consistent with the project? Apply the ref-python-coding-practices and ref-mcp-module-organization skills as relevant.
2. Correctness: Does the code do what it claims? Look for logic errors, incorrect assumptions, and edge cases.
3. Simplification and DRY: Can the code be simplified? Flag duplicated logic that should be shared, unnecessary abstraction, and over-engineering.
4. Code smells: Flag anything that makes a senior engineer pause and ask "why is it like that?" — long functions, deep nesting, magic numbers, complex conditions, oddly-shaped APIs (boolean-mode flags, stringly-typed parameters where a Literal or Enum belongs, long parameter lists), strange call syntax, dead or commented-out code, mixed abstraction levels within one function, mutable default arguments, broad except clauses, side effects in property getters or __init__, argument mutation, speculative generality, primitive obsession, feature envy, re-implementations of stdlib or library functionality, misleading names, inconsistent return types, sentinel returns where an exception belongs — or anything else that just looks off. This list is illustrative, not exhaustive; trust your judgment. Apply the general style rules in ref-python-coding-practices.
5. Security: Check for credential mishandling, session isolation issues, injection risks, and information disclosure — or anything else security-relevant. This list is illustrative, not exhaustive; trust your judgment. Flag any default or fallback ids — fully qualified ids arrive as explicit tool parameters and are validated by QualifiedSessionId.from_str, which raises rather than substituting a default.
6. Type safety: Flag any Any type hints, hasattr, or getattr usage without justification (per ref-python-coding-practices).
7. Closed-set dispatch exhaustiveness: For every dispatch on a Literal, Enum, or tuple of accepted values, verify a match + typing.assert_never(value) exhaustive form is used (or per-member metadata via __new__ for enums). Flag any silent if/elif/elif fall-through default branch on a closed set. See ref-python-coding-practices rule #18.
8. Suppression audit: Search the file for # pragma: no cover, # type: ignore, # noqa, # mypy: ignore-errors. For each occurrence, determine whether a design move would eliminate the need (factor a helper, narrow with cast, rewrite the API). Flag any suppression without an inline justification comment. Bare # type: ignore (no bracketed error code) is always a bug. See ref-python-coding-practices rule #19.
9. Docstrings: Apply the pydocs-improve skill to all functions and classes in the file. For any Pydantic schemas (StrictSchema / RedactableSchema subclasses), verify every field carries a PEP 257 trailing docstring — the project enforces this with tests/test_field_docs_contract.py; flag any reliance on Attributes: blocks as a documentation bug.
   • CLI help surface. For files under cli/_commands/ or cli/_help.py, the surfaced strings — command help= (via build_help), every click.option(help=...), and group docstrings used as help — are governed by ref-cli-help-standards, not pydocs. Apply cli-help-improve to that surface: verify the section contract, single-sourced OutputSpec, and plain-text (no-RST) rule. Internal docstrings on the same file still go through pydocs-improve.
10. Imports: Flag any unused imports. (run-precommit removes them via ruff; the reviewer flags them for awareness, does not edit.)
11. Logging: Apply the ref-logging-standards skill to review logging coverage and consistency.
12. Test coverage: Verify the file is covered by its corresponding test file at the project's 100% per-source-file target (see AGENTS.md and tests-improve). Flag any uncovered branch.
    • __init__.py files count. Every __init__.py — including ones that only define __all__ (even an empty __all__) or re-export from a sibling module — has its own dedicated test_init.py. The package surface is part of the project's API contract; an untested __init__.py is a silent-refactor hazard.
    • What the test_init.py must pin:
      • The exact set of names in __all__.
      • That every name in __all__ resolves on the package (hasattr(pkg, name)).
      • That each re-export is the same object as the internal definition (pkg.X is _module.X).
      • That no _-prefixed names leak into the public surface.
    • Canonical implementations: tests/config/schema/test_init.py, tests/config/test_init.py, tests/auth/middleware/test_init.py.
13. Output serialization: For any user-facing payload built in this file (MCP tool return dict or CLI OutputSpec field), apply ref-output-serialization-conventions to every string field — value vocabulary, casing, and known carve-outs.
14. Spelling: apply ref-python-coding-practices rule 8 to the whole file (identifiers and string literals included, not just docstrings); run uv run codespell <file> and flag what it reports.