The BATTLE HAMSTER Squeekchain Genesis upgrade is finished and playable, with the real wheel changing battles. One thing is still unproven: no moving wheel was seen live, because the hamster runs about 02:30–07:30 UTC, after this job's time limit.

The published site still shows the old version until the network publishes this dist/ to https://battle-hamster-squeekchain-genes.imd.site/. I left everything uncommitted in the working tree because .git/ is off limits, so the job's submission step has to pick it up. The added files total about 865 KB, well under the 8 MiB limit.

What was built

• Live status: every screen shows HARDWARE LIVE, SIMULATED, STALE or OFFLINE. A Wheel details dialog shows RPM, speed, distance, hardware status, provenance, freshness and last update.
• Battle effects while HARDWARE LIVE:
  • Wheel speed adds 0 to 12 energy per turn.
  • Verified distance adds 0 to 15% damage.
  • New settlements trigger an arena event: Block rush or Settlement tremor.
  • Any other status uses the original rules, and a lost signal is neutral on your next move.
• No built-in simulator: the game only shows what the API reports, and missing values read Unknown, never zero.
• Fixes:
  • Title text overlapping a sprite at 768px
  • Keyboard focus lost during turns and at battle end
  • The Leave dialog staying open after a battle ended
  • Another tab overwriting XP
• Added to the game:
  • A full battle history table
  • Upgraded stats shown in the loadout
  • Local vs on-chain labels on records

Real hardware evidence

• Live reading: the API reported hardware, online and module ready, with the wheel idle at 0 RPM.
• Distance, seen live: the real 6.35 km gave +15% damage. The same fixed-seed Attack dealt 21 damage instead of 18.
• Settlements, on real reads: the count rose from 270 to 274 during the job, and the detector fired on those real responses.
• RPM, not seen live: I ran real recorded hardware values through the rules instead. 5, 47 and 88 RPM gave 59, 63 and 67 energy after one Attack, against 55 neutral.

Test results

Not verified

• Browser to API directly: this machine's sandbox blocks Chromium from the internet, so the live browser run passed real API responses through Node. curl confirmed the API allows browser requests. Check that the published site shows HARDWARE LIVE.
• Safari and iPhone: WebKit can't launch here, and mobile checks were Chromium emulation.
• Real wallet approval: tested with a stand-in wallet provider only. No signature or transaction is ever requested.
• Screen readers, native zoom and RTL: not tested.

Blockers and questions for Squeekchain

• Blocked: on-chain match records and a shared leaderboard need an approved service or a new contract, so records stay local and are labelled that way.
• Worth asking before strong public claims: "hardware" is what the API reports, and the browser can't verify the device signature. The live signer is ed25519 where the docs' example shows trust-m, and the runs endpoint reports profile: staging.

While working I ran one git mv that staged a rename, and I undid it, so nothing i

Not delivered: artifacts/DESIGN-GUIDANCE-LICENSE.txt, artifacts/browser-results.json, artifacts/contrast.json, artifacts/genesis-readiness.md, artifacts/hardware-evidence.json, artifacts/live-hardware-390.png, artifacts/mutation-check.json, artifacts/validation.md. Under artifacts/, only this step's named outputs leave the machine.

Saved 5 verified findings: 2 high, 3 medium in [.imd-findings.json][findings]. Each includes exact source citations, a concrete reproduction, measured evidence, and a proposed fix. Application files remain unchanged.

• High — bug: A stored Block rush still adds damage after the hardware signal becomes offline, stale, or simulated: 23 damage instead of 18.
• High — fidelity: The Genesis readiness report and six linked evidence/result files are absent.
• Medium — bug: Concurrent battle completions in two tabs overwrite one match and 15 XP.
• Medium — bug: Failed statistics requests across the device’s noon rollover preserve an expired +15% distance bonus.
• Medium — accessibility: An AI victory behind the Leave dialog drops keyboard focus to BODY.

Verification passed: production build, TypeScript, 35 engine/API tests, 38 DOM tests, 39 Chromium browser checks, and the taste probe. Real API distance changed a seeded attack from 18 to 21 damage. WebKit could not launch because required system libraries were unavailable. Moving hardware and real wallet approval remain unverified.

The visual idea is a turn-based hamster arena carried by original pixel fighters in a torchlit burrow.

The requirement assessment follows the brief’s clauses, including repeated acceptance conditions:

Four of the five previous findings are fixed: settlement surge cancellation, expired distance bonuses, concurrent-tab saves, and result-screen focus.

One high finding remains: the Genesis readiness report and six linked evidence files are absent. Recorded in .imd-findings.json.

Validation passed: production build, TypeScript, 38 engine/API tests, 41 DOM/wallet tests, and 42 Chromium checks. Twelve concurrent-tab reproductions preserved both matches and 30 XP. WebKit could not launch because required libraries are unavailable.

Restored all six missing deliverables and .imd-responses.json, and fixed their Git exclusion. Game source and export remain unchanged.

Build, typecheck and 79 tests pass. Browser checks pass 40/42; two existing layout issues and verification gaps are documented in the Genesis readiness report.

The finding from last round is fixed and I found nothing new, so .imd-findings.json holds an empty list ({"findings":[]}). I changed no tracked files.

Last round's high finding: fixed

"The Genesis readiness report and linked evidence files are still absent."

• Files are there and in Git. All six files now exist, are tracked, and are not git-ignored: genesis-readiness.md (7.5 KB), hardware-evidence.json (338 KB), live-hardware-390.png, validation.md, browser-results.json and mutation-check.json.
• Links work. Every artifacts/ link in README.md and DESIGN.md points at a file that exists.
• The readiness report says what the brief asked. It gives a clear decision ("conditional"). It separates live, recorded and mocked evidence, and it lists six blockers.
• The hardware evidence holds raw API responses. These are /v1/status, /v1/measurement/stats and /v1/settlements bodies with their SHA-256 hashes.
• The screenshot matches the evidence. It shows HARDWARE LIVE, +15% damage and 6.35 km, the same as the recorded powerStripText.

Two layout failures the builder reported, checked and not raised

The builder's own browser suite records two layout failures (40/42 passed). I measured both in Chromium on the shipped dist/, with live API responses forwarded to the page:

• "HARDWARE LIVE" under "Wheel details" at 320px. With Arial-width fonts the badge ends at x=168.5 and the button starts at x=171.5, so they don't overlap (3px gap). At 375 the gap is 50px and at 390 it is 65px. The overlap only appears with the DejaVu Sans fallback in the builder's environment. That margin is thin, but I couldn't reproduce an overlap, so I didn't raise it.
• Hero text touching Pip at 1001 and 1024. The sentence's box reaches 12–25px into the sprite's box, but only over its transparent padding. On my screenshots the text doesn't touch any drawn pixel at either width.

Requirements

1. Use /v1/status, /v1/measurement/stats and /v1/settlements: met, src/ and the raw responses in hardware-evidence.json.
2. "Display RPM, distance, speed, hardware status, provenance, freshness and last update time": met, the power strip and Wheel details in src/components/SqueekPower.tsx.
3. "Real wheel activity increases arena energy": met. With recorded RPM inputs, energy after Attack is 59, 63 and 67 against 55 at neutral (genesis-readiness.md).
4. "Verified distance influences temporary battle modifiers": met. Live it gave +15% damage, and the battle log read "Squeek Power: +2 damage".
5. "New verified settlements can trigger arena events": met, the Settlement tremor when the count goes from 271 to 279.
6. "Activity must actually change battle mechanics": met, same evidence as items 3 and 4.
7. Separate HARDWARE LIVE, SIMULATED, STALE and OFFLINE, and never show simulated data as live: met. I saw OFFLINE myself when the API was unreachable; the 15,360-case signal test covers the rest.
8. Neutral modifiers when there is no live data: met, engine and DOM tests.
9. Three hamsters, AI turn-based battles, Attack, Defense and Special: met, browser suite.
10. HP, energy, XP, levels, upgrades, history, replay and leaderboard: met, browser suite and the empty leaderboard check.