feat(cli): add --category filter and --score output

[oekazuma]

Summary

Implements plan plans/018-cli-category-score-flags.md — two small quality-of-life flags:

• --category <cats> — restrict analysis to whole categories (seo | performance | correctness | security | architecture, comma-separated, case-insensitive). Previously "just check SEO" meant enumerating 30 rule ids with --rules. Applied as an intersection after --rules/--ignore/config selection, and threaded into the --baseline pass too, so category-scoped PR gates stay consistent. Health becomes the weighted average of the categories present (existing computeHealth behavior).
• --score — print only the combined Health score (one integer line on stdout, spinner and reporter output suppressed) for scripts and shell prompts. Exit-code semantics are unchanged, so --score --min-health 80 works as a quiet gate. Passing --reporter/--json alongside warns on stderr that reporter output is suppressed.

Notes

• Validation mirrors the --weights precedent (case-insensitive, de-duplicated, unknown categories are a fatal exit-2 error listing the known set). A value-less --category is ignored like a value-less --weights; a value with zero valid entries (e.g. --category ,) is a fatal error.
• resolve-args.ts's CATEGORIES list remains the single source both --weights and --category validate against.
• Out of scope (deliberately): a categories input on the MCP analyze tool — analyzeProject already accepts it, so that's a two-line follow-up if demand shows up.

Changes

• packages/cli/src/resolve-args.ts (parseCategories, --score wiring), index.ts (options + rule filtering + score-only output path), bin.ts (flags + help)
• 12 new test cases across resolve-args.test.ts and run.test.ts
• Docs: --category and --score sections in the CLI guide (en/ja)
• Changeset: minor for svelte-vitals

Verification

• pnpm typecheck, pnpm --filter svelte-vitals test (424 tests), pnpm lint — all green
• Live checks on the test fixture: --score prints exactly one integer line (exit 1 with findings present, exit gate honors --min-health); --category seo --json reports only SEO-prefixed rule ids

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features

  • Added a --category flag to limit analysis to selected rule categories.
  • Added a --score flag to output only the overall Health score for scripting.

• Bug Fixes

  • Improved filtering so category-based analysis respects existing rule selection and configuration.
  • Suppresses reporter output when score-only mode is used, with a warning if combined with other output formats.

• Documentation

  • Updated CLI docs in English and Japanese to describe the new flags and behavior.

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR adds --category and --score CLI flags to the analysis tool. --category restricts rule execution and Health score weighting to specified categories, intersecting with --rules/--ignore; --score prints only the combined Health score, suppressing reporter output. Includes parsing, validation, tests, and docs.

Changes

Category and Score CLI flags

Layer / File(s)	Summary
CLI help text and parser config packages/cli/src/bin.ts	Adds --category <cats> and --score help entries and extends mri boolean/string option lists.
Argument resolution and validation packages/cli/src/resolve-args.ts, packages/cli/test/resolve-args.test.ts	Adds parseCategories to validate/normalize category input, wires categories into resolved options, warns when --score conflicts with --json/--reporter, and adds corresponding tests.
Category filtering and score-only reporting packages/cli/src/index.ts, packages/cli/test/run.test.ts	Adds categories to RunOptions/AnalyzeOptions, filters selected rules by category in analyzeProject, threads categories into primary/baseline analysis, and reorganizes run() so --score short-circuits reporter output; adds tests for both flags.
Documentation and changeset .changeset/cli-category-score-flags.md, docs/src/content/docs/guides/cli.md, docs/src/content/docs/ja/guides/cli.md	Documents --category and --score behavior, health score weighting, and error handling in English and Japanese CLI references, with a changeset entry.

Estimated code review effort: 3 (Moderate) | ~25 minutes

Sequence Diagram(s)

sequenceDiagram
  participant User
  participant bin as CLI (bin.ts)
  participant resolveArgs as resolveArgs()
  participant run as run()
  participant analyzeProject as analyzeProject()
  participant Reporter

  User->>bin: run CLI with --category, --score
  bin->>resolveArgs: resolveArgs(argv)
  resolveArgs->>resolveArgs: parseCategories, validate categories
  resolveArgs->>resolveArgs: warn if score + json/reporter
  resolveArgs-->>bin: options { categories, score }
  bin->>run: run(options)
  run->>analyzeProject: analyzeProject(opts with categories)
  analyzeProject->>analyzeProject: filter rules by category
  analyzeProject-->>run: findings + Health score
  alt score enabled
    run-->>User: print numeric Health score to stdout
  else
    run->>Reporter: output findings via selected reporter
    Reporter-->>User: reporter output
  end

Loading

Possibly related PRs

• oekazuma/svelte-vitals#40: The --score flag directly relies on the combined Health scoring implementation introduced in this related PR.
• oekazuma/svelte-vitals#129: Both PRs modify analyzeProject's category/scoring behavior, with the related PR adding --weights support that interacts with category-based health scoring.
• oekazuma/svelte-vitals#42: The --category plumbing depends on the Category type model that this related PR modifies by narrowing valid categories.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely summarizes the main CLI change: adding --category filtering and --score output.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[oekazuma]

@coderabbitai review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

🧹 Nitpick comments (1)

packages/cli/test/resolve-args.test.ts (1)

152-198: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

LGTM!

Consider adding a test for mixed valid and invalid categories (e.g. --category seo,bogus) to confirm the error remains fatal even when some entries are valid.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/test/resolve-args.test.ts` around lines 152 - 198, Add a test in
resolve-args.test.ts around resolve() / --category handling to cover mixed valid
and invalid values like seo,bogus; the current category parsing tests only check
all-valid and all-invalid cases. Assert that resolve('--category', 'seo,bogus')
still returns options as null and emits the fatal unknown-category error in the
same way as the existing bogus case, even when some entries are valid.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@packages/cli/test/resolve-args.test.ts`:
- Around line 152-198: Add a test in resolve-args.test.ts around resolve() /
--category handling to cover mixed valid and invalid values like seo,bogus; the
current category parsing tests only check all-valid and all-invalid cases.
Assert that resolve('--category', 'seo,bogus') still returns options as null and
emits the fatal unknown-category error in the same way as the existing bogus
case, even when some entries are valid.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: d0d92188-214a-4bcf-a57c-71171b422036

📥 Commits

Reviewing files that changed from the base of the PR and between d56e1d4 and d9cb3ba.

📒 Files selected for processing (8)

• .changeset/cli-category-score-flags.md
• docs/src/content/docs/guides/cli.md
• docs/src/content/docs/ja/guides/cli.md
• packages/cli/src/bin.ts
• packages/cli/src/index.ts
• packages/cli/src/resolve-args.ts
• packages/cli/test/resolve-args.test.ts
• packages/cli/test/run.test.ts