Add pnpm and Vitest support for modern JavaScript stacks

[FlorianBruniaux]

Summary

RTK excels at git workflows (88.6% avg reduction verified) but lacks support for modern JavaScript package managers (pnpm) and test frameworks (Vitest), limiting adoption in T3 Stack, Turborepo, and Nx ecosystems.

Business Case

Market size:

• pnpm: 40% of npm downloads (rising rapidly)
• Vitest: Fastest-growing test framework (Vite ecosystem)
• T3 Stack: Dominant Next.js starter (create-t3-app)

Impact: Adding pnpm + Vitest support → 85% command coverage (vs current 40%)

Real-World Testing Results

Project: Méthode Aristote (production T3 Stack app)

• Next.js 15 + tRPC + Prisma + pnpm
• 43 Vitest tests, ~200 dependencies

Current state (v0.2.0):

• Git workflows: ✅ 88.6% avg reduction (validated)
• pnpm operations: ❌ Not supported (would save 80-90%)
• Vitest output: ❌ Not supported (would save ~90%)

Session impact:

• Baseline: ~11K tokens per dev session
• With RTK v0.2.0 (git only): ~5K tokens (55% reduction)
• With pnpm + Vitest: ~2.2K tokens (80% reduction) 🎯

Detailed Analysis

1. pnpm Support

Command: pnpm outdated

Current output (18.6K chars):

┌─────────────────────────┬─────────┬────────┬──────────┐
│ Package                 │ Current │ Wanted │ Latest   │
├─────────────────────────┼─────────┼────────┼──────────┤
│ @clerk/express          │ 1.7.53  │ 1.7.53 │ 1.7.65   │
│ Legend: <outdated> ...  │         │        │          │
└─────────────────────────┴─────────┴────────┴──────────┘

Proposed RTK output (~1.8K chars, 90% reduction):

@clerk/express: 1.7.53 → 1.7.65
@prisma/client: 6.5.0 → 6.6.1
next: 15.1.4 → 15.2.0

Superfluous content (60-70%):

• Box-drawing characters (┌─┬─┐)
• "Legend:" marketing text
• Full paths (node_modules/.pnpm/...)
• Empty rows

Implementation pseudocode:

fn filter_pnpm_outdated(output: &str) -> String {
    output.lines()
        .filter(|line| !line.contains("│") && !line.contains("┌"))
        .filter(|line| !line.starts_with("Legend:"))
        .map(|line| {
            let parts: Vec<&str> = line.split_whitespace().collect();
            if parts.len() >= 3 {
                format!("{}: {} → {}", parts[0], parts[1], parts[3])
            } else {
                String::new()
            }
        })
        .filter(|s| !s.is_empty())
        .collect::<Vec<_>>()
        .join("\n")
}

Commands to support:

• pnpm list --depth=0 (3.9K → 700 chars, 80% reduction)
• pnpm outdated (18.6K → 1.8K chars, 90% reduction)
• pnpm install <package> (filter progress bars, keep errors + summary)

---

2. Vitest Support

Command: pnpm test (43 passing tests)

Current output (10.5K chars):

 ✓ src/components/Button.test.tsx (3)
   [2m >[22m Button Component [2m >[22m renders correctly
   [2m >[22m Button Component [2m >[22m handles click events
   [32m ✓[39m [2msrc/utils/format.test.ts[22m (5)
[1m[46m PASS [49m[22m Waiting for file changes...

Proposed RTK output (~1K chars, 90% reduction):

PASS (43)
FAIL (0)
Time: 450ms

Superfluous content (85%):

• ANSI color codes: [1m[46m [32m [39m
• Checkmarks: ✓ → "PASS"
• Test hierarchy: [2m > [22m → flatten
• Full file paths → abbreviate

Implementation pseudocode:

fn filter_vitest_run(output: &str) -> String {
    let passing = count_pattern(output, "✓");
    let failing = count_pattern(output, "✗");
    let time = extract_regex(output, r"Time: (\d+ms)");

    let mut result = format!("PASS ({})\nFAIL ({})\n", passing, failing);

    if failing > 0 {
        result.push_str(&extract_failures(output));
    }

    result.push_str(&format!("Time: {}", time));
    result
}

Commands to support:

• vitest run (filter passing tests, keep failures)
• vitest --coverage (keep coverage %, remove tables)

---

Dev Effort Estimate

Total: ~5 days (1 week sprint)

Feature	Days	Implementation Complexity
pnpm list	1	Low (simple table parsing)
pnpm outdated	1	Low (same as list)
Vitest run	2	Medium (ANSI stripping + hierarchy flattening)
Testing + docs	1	-

ROI: 1 week dev → RTK becomes essential LLM dev tooling for millions of T3/Vite users

---

Testing Commitment

We can provide:

• ✅ Real-world benchmarks: 53KB test data (12 commands, production codebase)
• ✅ Implementation guide: Rust pseudocode ready to adapt
• ✅ Test cases: Edge cases discovered during production testing
• ✅ Documentation: Integration examples for Claude Code users
• ✅ Ongoing testing: Re-test after each PR on production codebase

---

Pattern Analysis (Cross-Tool Insights)

Common superfluous patterns across modern dev tools:

Pattern	% of Output	Tools	Fix
Box-drawing (┌─┐)	60-70%	pnpm, npm	Strip
ANSI codes ([32m)	20-30%	Vitest, Jest, colored	Strip
Marketing tips	5-15%	Prisma, pnpm	Remove
Verbose paths	10-20%	All	Abbreviate
Progress bars	5-10%	All	Remove

Insight: Modern CLI tools optimize for human readability but waste tokens for LLMs. Average reduction potential: 84% across unsupported tools.

---

Additional Context

Full test report: RTK-TEST-RESULTS-2026-01-28.md (available upon request)

Contributor guide: RTK-CONTRIBUTOR-GUIDE.md with complete Rust implementations

Project: Claude Code Ultimate Guide team (https://github.com/FlorianBruniaux/claude-code-ultimate-guide)

We're committed to testing and validating RTK improvements on real-world production codebases.

[pszymkowiak]

it's merged.