docs(triage): verification sweep — 2 of 8 first-batch tickets already fixed upstream

[kylehgc]

Verification sweep of every open first-batch adoption ticket against current develop, before adopting any of them. Triggered by the issue #3 discovery: upstream PR #2263 is still OPEN, but its bug was already fixed on develop by upstream 952245d — an open PR is not evidence its bug is unfixed.

Findings (one verification agent per ticket, code inspection + live repros on a develop-built binary)

Ticket	Upstream	Verdict
#1 cc-economics --audit	#2953	Still valid — feature absent from develop; its tracking.rs test-race fix strictly supersedes our partial local one. Moderate conflict risk (tracking.rs, main.rs)
#3 newline allow-bypass	#2263	Superseded by 952245d (split_for_permissions). Live repro: git log\ngit push --force-with-lease → ask; quoted newlines still allow. Recommend closing
#4 UTF-8 lossy decode	#2997	Still valid — all six map_while(Result::ok) sites present in stream.rs. Low conflict risk
#5 Windows code page	#2717	Still valid — no code-page decoding anywhere; adds encoding_rs. Touches high-traffic files: adopt soon
#6 runner stderr	#3029	Still valid — guard absent (runner.rs:114–130); also makes src/cmds/README.md:97's stderr-passthrough claim true. Low conflict risk
#7 git log --stat	#3028	Still valid — live repro: develop binary drops 8 of 30 commits, parses diffstats as headers (git.rs:455/580). Low conflict risk
#8 vitest cluster	#2982+#1497+#1922	All three still valid, but rtk-ai#2982/rtk-ai#1497 will not cherry-pick cleanly (develop's vitest refactors) — plan hand-port, order 1922 → 2982 → 1497
#9 npx passthrough	#937	Superseded — develop fixed it better via npm_cmd::exec + regression test citing the same bug. Recommend closing

Doc changes

• Triage first-batch table: rank 1 marked adopted (PR fix(hook): emit ask decision for Claude rewrites (adopt upstream #3031) #10); ranks 2/8 struck out as superseded with evidence; remaining ranks annotated with verification findings and conflict risk.
• New "Verification sweep" section recording the method and the process lesson: staleness is checked at adoption time, not triage time — repro-before-adopt is what catches it.
• Replacement tickets promoted from runners-up: #3027 and #2965 (filed as new issues).

🤖 Generated with Claude Code