coding

/dotclaude:coding — coding hygiene elicitation

You are eliciting the project-specific coding-discipline intent for the user's project: the right file-size ceiling and the bug classes a reviewer should specifically watch for. The output is ONE thin artifact — .claude/dotclaude/code-anti-patterns.md — plus a handful of keys in dotclaude.yml. You do not author an agent, a rule file, or a hook: the consumed code-review agent (shipped by the plugin — see agents/code-review.md) already reads dotclaude.yml's artifacts.code-anti-patterns path at runtime, falls back to its own generic methodology when the artifact is absent, and grades new changes against each entry in addition to the generic pass when it's present (full contract: ../../docs/artifact-contract.md).

This is the layer the user wants set up FIRST on any project — the elicitation is cheap, and the consumed code-review agent is useful even with an empty artifact. The companion /dotclaude:testing setup layers on top for test discipline, and the separate dotclaude-design plugin covers design / UX / a11y discipline for projects with a human-facing surface.

Phase 1 — Read the project's code shape

Before any question:

1. Stack signal — read whichever exists:

   cat package.json 2>/dev/null
   cat Cargo.toml 2>/dev/null
   cat pyproject.toml 2>/dev/null
   cat go.mod 2>/dev/null
   

   The dependencies + scripts disclose: language, framework, runtime, test runner, lint setup.

2. File-size distribution — find the worst offenders:

   find . -path ./node_modules -prune -o -path ./.git -prune -o \
     \( -name "*.ts" -o -name "*.tsx" -o -name "*.js" -o -name "*.py" \
        -o -name "*.rs" -o -name "*.go" -o -name "*.swift" -o -name "*.kt" \) \
     -print 2>/dev/null | xargs wc -l 2>/dev/null | sort -rn | head -20
   

   The top of this list calibrates the right ceiling: pick at the 95th percentile of healthy files + a buffer. A 1500-line outlier doesn't justify a 1500 LOC ceiling — it justifies decomposing the outlier.