Write and audit code against house coding conventions with rule-cited findings.
npx skills add https://github.com/wkentaro/skills --skill writing-code-wkentaro
Please help me install this Agent Skill. Skill: writing-code Source: https://github.com/wkentaro/skills/tree/main/skills/in-progress/writing-code Command: npx skills add https://github.com/wkentaro/skills --skill writing-code-wkentaro
file:line, the rule's section and bold lead-in, and a concrete edit, or report exactly No findings. - Cross-language application: Apply the structural rules beyond Python (scoping, extraction, imports, annotations, real dependencies over mocks) while dropping Python-only spellings like the _ private prefix. - Use Case: While reviewing a pull request, paste the changed test file and receive a numbered findings list flagging the test class, the deferred import, the bare boolean argument, and the missing -> None return annotation, each with its fix. ## Quick Start Use the writing-code skill to audit this change against the house coding conventions and list any rules it breaks.High-intent search queries and answers about installing and using this skill.
Paste the changed code and ask for an audit. The Skill checks every rule and returns a numbered list of only actual violations, each with file:line, the rule's section and bold lead-in quoted verbatim, and a concrete edit. If nothing breaks, it reports exactly "No findings."
Start function names with a verb naming what they do, such as compute_mask_iou or round_bbox_to_int. Non-verb forms are allowed only for predicates (is_*, has_*), classmethod constructors (from_*), and conversions (to_dict). Noun-only names like mask_iou are rejected.
Yes, the structural rules carry over: scoping, extraction, thin entry points, imports, naming, comments, annotations, and real dependencies over mocks. Python-only spellings like the underscore private prefix are dropped, while goals like keyword arguments translate to options objects or named struct fields.
Prefer real dependencies wherever cheap: existing fixtures, ephemeral subprocesses, dockerized services, or in-memory engines. Reserve mocks for paid third-party APIs, irreversible side effects like payments or emails, and services with no offline mode, since mocks keep passing after the real API changes.
The Skill states plainly that no rule covers the case instead of inventing a preference. This applies to style calls, audits, and every language, and it outranks giving a complete-sounding answer.
Run the repository's configured formatters, linters, and type checks before auditing. The audit does not repeat their findings or assume rules the repository has not enabled; it covers only judgment calls that tooling cannot prove.
otomattyCreate and update GitHub pull requests with automated checks and template descriptions.
basedgod55hjlUpdates an existing pull request with new commits and revised metadata.
Ramprasad4121Prepare codebases for security review with static analysis and documentation.
zhizhunbaoConfigure Ruff, Prettier, ESLint, and mypy with pre-commit hooks for Python and TypeScript.