feat: add yarn workspace command with smart filter routing

[denneulin]

Summary

Add rtk yarn workspace <pkg> [run] <script> command that strips yarn boilerplate and routes script output
through existing specialized filters for maximum token savings.

• Boilerplate stripping: Removes YN-prefixed info lines, resolution/fetch/link progress, yarn classic
  headers (yarn run v1.x, Done in Xs, info lines). Returns ok ✓ when output is pure boilerplate.
• Smart filter routing: Routes workspace scripts to existing RTK filters based on script name:
  • vitest → vitest parser, tsc/typecheck → tsc filter, lint/lint:* → generic lint filter,
    prettier/format → prettier filter, test/test:* → test runner summary
  • Prefix exclusions prevent misrouting (test:e2e, test:playwright, test:cypress, lint:fix →
    passthrough)
• Graceful degradation: catch_unwind on filter panics, empty-output guard, fallback to stripped output
  with stderr warning
• Early-exit on failure: When yarn itself fails (non-zero exit), outputs stripped stderr/stdout directly
  without routing through filters (prevents error masking)

Changes

File	Change
src/yarn_cmd.rs	New module (704 lines): filter_yarn_output(), route_script(), apply_filter(),
run() + 25 tests
src/main.rs	Register Commands::Yarn with trailing_var_arg + 5 Clap parsing tests
src/lint_cmd.rs	filter_generic_lint: fn → pub(crate)
src/runner.rs	extract_test_summary: fn → pub(crate)
src/tsc_cmd.rs	filter_tsc_output: fn → pub(crate)

Test plan

• cargo fmt --all --check passes
• cargo clippy --all-targets passes
• cargo test --all passes (30+ new tests)
• Manual: rtk yarn workspace <pkg> run vitest on a real yarn workspace project
• Manual: rtk yarn workspace <pkg> run typecheck with typecheck not existing

[pszymkowiak]

Review: PR #226

Thanks for this PR! The FilterRoute enum and early-exit on failure are well designed. I tested against a real Yarn 1.x workspace.

Bugs found

1. Non-workspace commands broken — rtk yarn install, add, list, why all fail with usage error. Fix: passthrough for non-workspace args (like pnpm_cmd.rs).

2. Boilerplate leaks — warning ..., yarn workspace v1.22.22, $ vitest run not stripped. Regex ^yarn run v\d misses yarn workspace v\d.

3. Vitest routing broken — route_script("test") → generic TestRunner → OUTPUT (last 5 lines) fallback. Same issue as #232: script name doesn't identify the tool.

4. Double stderr — Line 290 prints filtered output, line 293 prints raw stderr again on top.

5. Subcommand interception — rtk yarn workspace app add lodash runs yarn run add instead of yarn add. Need a denylist of native yarn subcommands.

Minor

• catch_unwind → use Result-based fallback (RTK convention)
• lazy_static! inside function → move to module level
• Token savings test at 40% → project standard is 60%
• Lots of overlap with your feat(pnpm): add rtk pnpm [run] <script> with smart filter routing #232 — consider sharing patterns

Architecture is solid, just needs these fixes. Happy to re-review!

[ThomasHoussin]

Hi @denneulin - I opened #317 as a clean rewrite of your yarn workspace support, rebased on #241 (rtk rewrite architecture). Your original implementation was the foundation - I integrated all 5 review fixes and extended it with non-workspace filter routing (rtk yarn test/lint route to specialized filters), compound script routing (test:lib, lint:check route by prefix with safe fallback), and registry integration for rtk rewrite/discover. The PR depends on #241 merging first. Happy to coordinate if you want to review or have feedback.

[FlorianBruniaux]

Hi @ThomasHoussin, @dlmw, @denneulin! You've each opened yarn-related PRs (#317, #272, #226) that all create yarn_cmd.rs — they'll conflict with each other. Could you coordinate? @ThomasHoussin's #317 is the most comprehensive, so we're thinking of using it as the base. @dlmw, @denneulin — could you check if your specific additions (vitest support, workspace support) could be contributed to #317 instead, or if #317 already covers them?

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[pszymkowiak]

Hi! Two things needed before we can review:

1. Retarget to develop — this PR targets master, but all PRs should target develop. You can change the base branch in the PR settings (right sidebar).
2. Sign the CLA — if not already done, please sign at https://cla-assistant.io/rtk-ai/rtk

Thanks!

[aeppling]

Hey

We are cleaning up the codebase and improving the project structure for better onboarding. As part of this effort, PR #826 reorganizes src/ from a flat layout into subfolders.

No logic changes — only file moves and import path updates.

What you need to do

Rebase your branch on develop when receiving this comment:

git fetch origin && git rebase origin/develop

Git detects renames automatically. If you get import conflicts, update the paths:

use crate::git;        // now: use crate::cmds::git::git;
use crate::tracking;   // now: use crate::core::tracking;
use crate::config;     // now: use crate::core::config;
use crate::init;       // now: use crate::hooks::init;
use crate::gain;       // now: use crate::analytics::gain;

Need help rebasing? Tag @aeppling

[ThomasHoussin]

I rebased #317, which I think covers #226 and #272. I think we could close both #226 and #272 and adjust #317 if necessary.

[pszymkowiak]

Hi @denneulin — closing per @ThomasHoussin's request on 2026-03-30:
his #317 rebased your work onto develop (post-#826 layout refactor),
integrated the 5 review fixes from above, and extended to non-workspace
yarn commands. He explicitly asked to close #226 and #272 in favor of
his consolidated branch.

Your original implementation was the foundation — the FilterRoute
enum and the early-exit-on-failure design carried over to #317. Thanks
for the work that started the conversation.

If you want to follow the merge, #317 is the active thread.

Closing as superseded by #317.