[deadp]

Added blind8a-d wrapper tests (seeds 1-8, 9-16, 17-24, 25-30) plus a shared blind8.ts helper that prints one compact summary block; no env vars, blind.ts unchanged.
vitest.config.ts collects blind8* only when named on the command line, so npm test is unchanged.
Added a PR 20 measurement note to docs/BACKLOG.md.
npm test (388 passed) and typecheck pass; the blind8 files were not run, as requested. Bash worked, but git --version was refused for approval so its output was not seen.

Queued from the dashboard (item 582). The request:

Goal (proposal 8, measure PR 20 on main): PR 20 (chapter 2 goal: sell 3 item kinds) is merged. Measure 30 seeds with the blind sim via LONGRUN, which only runs allow-listed commands (npx vitest run ), takes no env vars, returns only the stdout tail, and has a 90 minute limit. A 15-seed run (packages/game/scripts/blind30b.test.ts) hit that limit, so the sim must be split into chunks of at most 8 seeds.
Do: read packages/game/scripts/blind30a.test.ts (if present) and blind30b.test.ts to see how seeds, the sim loop and the summary are set up. Add four wrapper test files in packages/game/scripts/ that reuse the same code with fixed seed ranges and no env vars: blind8a (seeds 1-8), blind8b (9-16), blind8c (17-24), blind8d (25-30). Each prints at the end ONE compact summary block: seeds run, rare fleece reached by season 27 (count), median season of chapter 3 arrival (and min/max), reached the ending (count), plus the per-seed chapter 3 season list on one line. They must be excluded from the normal npm test (follow how blind30b is excluded) so CI is unchanged. Do not run them (too long); run only npm test and typecheck. Do not change any game, balance or config code.
Also add a short note to docs/BACKLOG.md: PR 20 merged 27-10, baseline chapter 3 median season 16-17 and rare fleece 2 of 30 by season 27, target chapter 3 about 11 and rare 15 of 30; revert PR 20 if the 30-seed numbers show no gain.
First command: git --version; say in your result whether Bash worked. Done

Opened by the queue runner on LXC 113. Merged once CI checks pass and the reviewer's verdict is recorded, unless it touches a hard line (qrunner/hardlines.toml): then it waits for Joshua's yes on the dashboard.

🤖 Generated with Claude Code