chore(scripts): remove 6 verified-dead per-package scripts (#10200)

[lalalune]

What & why (#10200 de-larp slice)

Fresh audit of scripts/, packages/scripts/, and the inventory-tool-uncovered packages/*/scripts + plugins/*/scripts for the issue's three de-larp axes. Pure subtraction — removes 6 airtight verified-dead scripts.

Axis results

Axis	Finding
Larp catch-verifiers (main().catch(console.error) → logs but exits 0)	0 remaining — already fixed by #11548; re-verified on develop tip.
|| true (42 occurrences)	All legitimate best-effort (command -v probes, idempotent sed/cp, grep -c no-match). No swallowed verify assertion. Deferred per #11548's per-callsite note.
Trivially-safe dupes	check-i18n.mjs / check-secret-hygiene.mjs exist in both packages/scripts/ and packages/app-core/scripts/ but have diverged (in-repo ../.. vs fork process.cwd() root resolution — intentionally separate). NOT safe to dedupe.
Dead scripts	6 removed (below).

Dead scripts removed — each 0 references tree-wide (proven git grep -l <basename>, none *.test.*, none imported)

• plugins/plugin-training/scripts/gepa-view-switching.ts — completed GEPA view-switching experiment; package.json wires gepa:view-context/verify:view-switching (kept) but not this. Same class as GEPA one-offs deleted in chore(scripts): #10200 slice — delete 15 verified-dead root scripts, fail loudly on script rejections #11548.
• plugins/plugin-training/scripts/multishot-view-switching.ts — completed multi-shot-demo finding experiment; unwired.
• packages/app/scripts/android-webview-attach-probe.mjs — one-time Playwright-Android-attach "de-risk probe"; real harness now exists.
• packages/benchmarks/social-alpha/trenches-chat-dataset/scripts/monitor_progress.sh — manual watch-loop helper; README pipeline is only build_dataset.ts + fetch_price_history.ts.
• packages/benchmarks/social-alpha/trenches-chat-dataset/scripts/calculate_success_metrics.ts — one-time metric analysis; not in README pipeline, not imported by build_dataset.ts.
• packages/benchmarks/social-alpha/trenches-chat-dataset/scripts/calculate_realistic_success_metrics.ts — iterated near-dup of the above.

Verification

• Pre-deletion git grep -l "<basename>" returns only the file itself for all 6.
• Post-deletion: no dangling references anywhere; no tsconfig files array lists any of them.
• All are standalone entrypoints with no importers → deletion cannot affect any package build or typecheck.
• plugin-training retains its wired harnesses; trenches README pipeline untouched.

Honest scope note

Prior #10200 slices (#10479, #10681, #11548, #11367) already exhausted the safe dead-script surface in scripts/ + packages/scripts/, fixed all silent-exit 0 handlers, and corrected the inventory orphan model. Remaining per-package orphans are overwhelmingly manually-run operator/dev/QA/migration tools (#11367 lesson: hand-run entrypoints are false orphans) and are deliberately left in place. The 6 here are the airtight completed-throwaway subset.

Evidence: .github/issue-evidence/10200-scripts-delarp-slice/README.md

App/native/screenshot evidence: N/A — repo support-script deletion only, no UI/runtime/native surface.

🤖 Generated with Claude Code

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: aa682527-e6c0-4424-a5f0-4dca974b92f1

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch chore/10200-scripts-delarp-slice2

---

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.}

[lalalune]

Same-account review note, not an approval.

No findings. I checked the deleted helpers against the package-local docs and the live script surfaces:

• plugins/plugin-training/CLAUDE.md describes the runtime/plugin surface and CLI scripts; the PR only removes two standalone experiment helpers. plugins/plugin-training/package.json still wires the kept scripts/gepa-view-context.ts and scripts/verify-view-switching.ts harnesses, not the deleted gepa-view-switching.ts / multishot-view-switching.ts.
• packages/benchmarks/social-alpha/CLAUDE.md and trenches-chat-dataset/README.md document the benchmark/dataset path. The dataset README pipeline names only scripts/build_dataset.ts and scripts/fetch_price_history.ts, both still present.
• packages/app/scripts/android-webview-attach-probe.mjs is not referenced by app scripts after deletion; the remaining Android scripts are the real capture/e2e/install helpers.

Local Windows validation:

• Refreshed origin/develop; git diff --name-status origin/develop...review/pr-11627 shows only the evidence README plus the six deleted scripts.
• git diff --check origin/develop...review/pr-11627 -> clean.
• Exact deleted-path scan across manifests/docs/tsconfigs (excluding the new evidence README) found no references.
• Basename scans on the PR branch found no matches outside the evidence README for the app/plugin-training names; benchmark helper names had no matches in packages/benchmarks/social-alpha, benchmark registry/tests, or root manifest.
• git ls-tree confirms retained live helpers: gepa-view-context.ts, verify-view-switching.ts, verify-view-switching.grid.test.ts, build_dataset.ts, and fetch_price_history.ts.

No runtime/UI/native surface changes here, so screenshots/video are N/A for my review.

[github-actions]

LifeOps Benchmark — eliza

Run ID: lifeops-eliza-28624701941

LifeOps Benchmark

Model: gemma-4-31b
Judge: claude-opus-4-7
Scenarios: 25
pass@1: 0.000
pass@k: 0.000
Total cost: $0.0000

Full artifacts: see the lifeops-run-eliza-28624701941 upload on this run.

[claude]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.