The rule (drivertable::install_hold_keys): a driver install on a card the app reads as external (the eGPU kind from update-return-21b) stops that card's worker before the installer starts, through the cards path with the restore choice kept (the --cards-off shape); the vendor's cards inside the case and every other vendor's card keep mining. The row says so before the click and while it runs (the display reset can take the machine for a minute and may need a restart; save your work first). When the installer ends the held card goes back as it was; a card the install took away comes back when the card does (driver_release_held on the detection that lists it again). The app never restarts the machine.
Tests: the known-failed rule test first (an install on an eGPU card with the worker still running held nothing: red on build-2, then green), the view test for the two sentences; box gate 258 + 33 + 8 (test --release on build-2). Mock scenarios drivers-egpu and drivers-egpu-running; captures 13 to 16 (light and dark). Branch rebased onto release-0.3.21, which already carries the driver-check commits; the earlier tip is kept as driver-check-0320.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
tools/scene/parity.mjs serves site/ (tools/site-serve.mjs) and the app's UI (tools/ui-mock/server.mjs) on a build box, answers
every page's /api/live from scene/fixtures/live-2026-10-07.json, freezes the clock at the fixture's instant (Date, timers,
requestAnimationFrame) and reads each scene canvas at 900 by 420 px at the first push and the next two polls: the read is the
push's own synchronous paint, so a frame depends on the fake time alone, never on how long a fetch took. Known-failed first: the
app with --included moved by one unit must differ from /live at every instant (it does: 1,549 / 1,282 / 1,157 px). Then home
fold = /live, app Inspect = /live, app with this machine's key = /live with the same key (the overlay is the same picture), and
the app's overlay frame differs from its base frame (the overlay is drawn). The compact card is rendered and reported, not
compared. A RED line names the differing pixels and their box. tools/scene/parity-remote.sh carries the files to the box
(~/.config/igneum/build-server-2 by default) under this lane's prefix and runs it there; a plain CI runner with no box and no
Playwright prints a skip line. One new line in tools/ci/pre-push.sh.
First run on build-2 at 15:4x UK, release-0.3.21 tree plus scene-parity: every comparison equal at T+0, T+2 and T+4 s.
Found on the way and fixed in the harness, not the renderer: lane order keeps the history of earlier layouts (the scene must be
at the compared size before its first layout); the app's recorded mock card is a real devnet key (cleared for the base case).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 4102c935e518e83eba39d880daad1a57b77c3bf8)
The blank /live (the project lead, 14:3x UK): IgneumDag 2.0.2 painted only from a requestAnimationFrame loop gated on document.hidden and
the IntersectionObserver, so a page that loaded with document.hidden true (the desktop pane, a background tab) and whose embedder
never fired visibilitychange took every push, reported live and never drew a frame. 2.0.3 paints the current picture synchronously
on push, size and theme change; the motion loop alone waits for a visible document and an intersecting canvas. Confirmed headless
on build-2 against the live feed: hidden document 0 lit pixels before, 110,007 after; the never-intersecting observer repaints
on every push. Known-failed test tools/scene/paint-test.cjs (the 2.0.2 renderer draws nothing in the same world).
The second 2.0.3 change: the phone rule (30 s window, four lanes) keys on the viewport width, not the canvas width; a 640 px hero
on a 1,440 px laptop was rendering as a phone while the app's card beside it was not.
scene/ is the one source: live-dag.js, proof-core.js, tokens.css (the fourteen palette tokens, the brand package's values, dark and
light), feed-contract.md and .json (one JSON shape for the observer's /api/live and the app's api/live), a recorded reply as the
fixture. tools/scene/sync.mjs writes the copies and the scene-tokens block into site.css and app.css; --check is the gate line
(byte-equal scripts, an equal block, the names defined nowhere else, a print block excepted), --self-test fails five known cases
first. The site's token definitions move out of the package's :root line into the block; no value changes on the site.
tools/scene/feed-contract.mjs validates a reply against the key lists; its test refuses a miner rewritten to "you", a float now,
a stray key. Three new lines in tools/ci/pre-push.sh. The app side (branch scene-parity, for 0.3.21) takes the same folder.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit f774353461)
docs/plans/miner-faults.md: MF-1 to MF-7, each with its rule, test and gate line.
- MF-1/MF-2: a worker starts and is judged only when the node is READY (synced and igneum_getExecStatus reports an
executed tip; execrpc::probe every 5 s off the engine thread); the node watchdog never counts the catch-up (settled
once read synced; 30 min cap before that; any RPC answer is a sign of life); the watchdog restarts on a ladder 10 s,
30 s, 2 min, 5 min, then every 5 min for ever (watchdog::RETRY_LADDER_S); the faulted state and the one-restart
budget are gone (tools/ci/permanent-fault-check.sh in the gate); a node-caused restart resets the ladder at sync.
- MF-3: the hot-plug pass starts a recovered or revived card's worker (unchanged rule, now in the register).
- MF-4: the status clock starts at ready (program loaded), loading bounded by 300 s; a self-test failure holds the
card 30 min with the reason on its row, released on a driver change; a crash loop climbs the ladder; the pack is
exported once a minute for every card (a refused pack forces one).
- MF-5: the app reads template_wait=, template_ms=, identities_active= from the 0.3.20 miner's STATUS; waiting on
the node is never the card's fault; the row says node slow; every node-wait label clears on the first rate.
- MF-6: a miners hold belongs to the job that took it and releases when that job is gone or at its own cap.
- MF-7: the engine owns every igneum-miner it started: an untracked one on this engine's node RPC is killed at start,
after every stop and every minute, one line and one fault report per kill; a restart kills the old process first.
- Every fault line posts one FAULT line to the log intake (label fault-<id8>, app and node version, 60/h cap).
- The signed cards job kind (per card enabled, identities, power_pct; refused for a card the machine lacks; applied
through the app's own card path, persisted, read back): packaging/ota/publish-jobs.sh add --kind cards.
- LG-4 as a job: relay/playbooks/first-share.ps1 and tools/fleet/first-share-gate.mjs (no Windows box yet).
- tools/reliability: the fault injector with one step per class (catch-up, card-appears, own-restart, zero-ladder,
no-status, node-silent, one-card-fails, orphan-miner); fake-worker.mjs lists devices and fails self-tests on command.
- master's build tooling (97255a4e) and release-0.3.20's igneum-pow taken into the worktree for the box routes.
Box: app 198 + 27 + 8 tests green on igneum-build-2; the tree gate green (33 checks).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
NVIDIA GeForce Game Ready 617.42 WHQL (6 October 2026): us.download.nvidia.com/Windows/617.42/..., 990,853,168
bytes, sha256 f115c927..., subject CN=NVIDIA Corporation (DigiCert G4). AMD Software Adrenalin 26.9.2 WHQL
(29 September 2026, the win11-b build): drivers.amd.com/drivers/whql-amd-software-adrenalin-edition-26.9.2-win11-b.exe
(an amd.com Referer required), 1,000,800,840 bytes, sha256 593c1d73..., subject CN=Advanced Micro Devices (Sectigo).
Read on the box: curl, sha256sum, the PKCS7 out of the PE security directory through openssl pkcs7 -print_certs.
The table's placeholders are gone: every row installs. The drivertable test sample, the mock and the view test name
the real NVIDIA release. detect.rs:990 carried a #[test] above the doc comment of the Intel test from the cherry-pick,
the test build's one warning ("duplicated attribute"): removed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 80787ea024)
the project lead: can the drivers be packaged with the miner, for all cards, with the system knowing which to install if not
present. Not bundled: detected and installed on one click. A per-vendor table rides the signed manifest (drivers.json:
min_version, the version on offer, the vendor's URL, size, sha256 and its source page, the silent arguments, the
restart exit codes, the Authenticode signer, the Linux and HiveOS package), validated by the signer and the app alike
(src/drivertable.rs, shared), written to <app data>/drivers.json by ota.rs. Each card's driver version is read at
every detection (nvidia-smi's driver_version; Windows' DriverVersion for AMD and Intel) and compared; a missing or
old driver puts the offer on the card's row, on the dashboard and on the first-run list. The click downloads with
curl (resume), checks size, sha256 and the Authenticode subject, runs the installer through one elevated prompt
(platform::elevated_command, the PC 1 driver job's shape), reports restart required with a Restart now button, and
never restarts by itself; the miners keep mining. macOS: no step; Linux and HiveOS: the package line. Dry run through
IGNEUM_DRIVER_DRY_RUN or the table. Tests: the table, the versions, the offers per tier, the exit codes, the
Authenticode verdicts, the download against a mocked vendor server on 127.0.0.1, the UI's strip per state; the mock's
drivers scenarios; captures light and dark in docs/plans/driver-check-shots. publish-manifest.sh --drivers carries the
table. Also the doubled #[test] in detect.rs from the cherry-pick.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 3f0ef65786)
The red job's condition drops the master/release-* clause; record() keeps GITHUB_ACTOR and the head commit's author name (RED_WATCH_AUTHOR from the workflow) and formatLine prints "pushed by <actor> (commit by <author>)" before the failed job and step. The self-test covers a feature-branch run and the author in the line. Same updates channel, same box file, same one-line-per-run idempotence.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The class (7 October 2026, 11:26 to 12:47 UK): three fork-gate summaries on ca3-v4-node carried the miners' vote-key hashes under "key" and eight CI runs went red on "no secret file names and no 64-hex secrets in the tree" while the pushing lanes saw nothing: the light gate ran only conflict markers and Windows paths.
- tools/ci/pre-push.sh: never_push_checks() (identity grep, no-secrets) runs on every ref from --hook, and inside the full gate where the identity grep already sat; the self-test asserts the wiring; the light gate is about 20 s on the Mac.
- infra/fast-time/lib/redact-keys.mjs: writeSummary() shortens every 64-hex value under a key-shaped field to 8 hex and an ellipsis and refuses a text the no-secrets rule would flag (line named); --self-test and --check; the gate runs the self-test. fork-gate.mjs adopts it on ca3-v4-node.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
the project lead, 7 October 2026 15:02 UK: build-1 at load 139 / 114 / 90 with both slots held and a 1 h 40 min queue while build-2 read 4.5 with
free slots, because the class router pinned each class to its box. Now lib.sh bs_route_spill reads the preferred box (free slots,
1-minute load) with one ssh and hands the job to the other box when the preferred one has no free slot or sits above load 64 and the
other qualifies; neither qualifying queues on the class's own box. The decision travels as BR_ROUTE_* into the JSONL "route" object
for the dashboard. build-2's slots file reads 3; everything on box 2 runs at nice 10 / 32 cores / -j 32, and a bounded run takes
the band its slot owns so three never share a core. Self-test tools/ci/route-spill-check.sh (thirteen cases) in the gate.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
docs/community/discord/structure.md: every channel with topic, pinned first post, slowmode and thread settings; the
AutoMod rules; the rules screen; the server guide to-dos and the pre-join question aligned with the map; the two bot
specs; the office-hour template; what needs the founder. Every row carries "set / read back"; all wait on the Discord
sign-in (the session expired; nothing typed).
tools/community/discord-hooks.mjs: `feed` (hourly: height, hash rate, keys, the last lock from /api/live), milestones
posted once each, the daily hash-origin field, `--via updates`, optional DISCORD_WEBHOOK_FEED, the tick wiring. Six
tests, 37 pass. One live post through the updates webhook as the proof of the path (feed:2026-10-07:11).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Intel's compiler turns rotate(x, (0u - n) & 31u) into a rotate LEFT by n: lane 0's register trace on the B580 diverged
at instruction 6 of iteration 0 (rotr) and nowhere before, in both exchange modes, with every other family and the
dataset kernels bit-exact. proto-opencl/intel_rotr.h rewrites the one helper line when the device's vendor or
platform string holds Intel (host.c's buildProgram and the prepare path), no other vendor sees a change, no pack or
consensus text moves. proto-opencl/test_intel_rotr.c (the pre-push gate runs it) feeds the line through the rewrite
under Intel, AMD and NVIDIA strings and asserts the outputs.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 26e135a362718a68a842da59080b43e93afd9dc2)
The dashboard lane, 7 October 2026 10:39Z: both slots of build-1 flock-held and EMPTY while two suites ran. Cause: a run from a
worktree without last night's append-mode fix opens a busy sibling's slot file with > on every probe. The holder now keeps its own
line: a keeper re-writes it within BR_KEEP_S (20 s) whenever the file is empty, until release; remote-run.sh --self-test-keeper
(in the gate) truncates a held line and sees it return, and sees nothing written after release; live on build-1 at 11:19Z (the
line came back in 25 s). The first version deadlocked the runner's bare wait with the keeper (build-2's first run hung 15 min
after its test passed): the keeper stops before the wait. The two running suites' -j 90 came from explicit --jobs 90: a bounded
class now clamps it to its cap with a log line (pass --priority gate for the full set). run-from-mac.sh --box N: the host file
was suffixed twice (build-server-2-2) and every box's mirror would have shared one remote name; one remote per box (build-N).
build-2's first green run: a suite at nice 10 on 32 cores, jobs 32, 986 s cold.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The owner, 12:4x UK: "some things look different on the live site vs the one I sent you, match it 1:1". site/site.css is
now the package's design system as delivered (tokens, type, chrome, components, breakpoints) with self-hosted fonts, the
two text tokens for contrast on light surfaces, and a compatibility layer for the data pages' vocabulary. The header
and footer are the package's (status bar, logo with the ember stop, the six items, the Discover dropdown, the theme
toggle, the Download button, the burger and its sheet; the footer grid with the three icon links, Run, Read, Explore
and the legal line). The home, miner and wallet pages carry the package's markup on our content; the live page, the
explorer, the litepaper, the faucet, the MetaMask page and the generated pages take its values in their own blocks.
The theme is always set (stored, else the OS), so the package's light theme applies. Side-by-side sheets before and
after at 1440 px from tools/site-redesign/sheets.mjs; no sample data, no review chrome, the hero loop continuous.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
tools/heat-gate.mjs reads the HEAT lines of an app log and passes a hold within 1 degree for 4 hours with the hash
following the slice; its self-test fires on a known hold, a drift, a hash through the rest, a short log, an empty log
and a log without readings, and sits on the one gate beside heat-region.test.mjs. docs/plans/ember-heat.md carries
the words, the loop, the sources, the per-tier table and the runbook for the project lead's desk (this lane never touches PC 1).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 0d5fc6d258dee4774ebe7ea760736c9f05026d06)
First run gains step 2 of 3 (the region list, the price per kWh prefilled from the public table and typed, never
fetched, the restricted line for Russia's regions and China from future.md 4.4 and 8.3 with the standing sentence);
Settings gains Electricity and Heat mode; the Cards strip and every row's Ember line show the duty and the heat in
watts; a resting card, the pill and the big button say heat mode; Earnings gains the miner's cost per MH/s-hour at
their price beside the bench log's measured rental rate with the verdict. The money symbol follows the region.
heat-region.test.mjs checks the Russian entry, the bench-log rate and the words; the mock gains heat and heat-rest.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 8099bbfd46)
publish.mjs packs the fifteen served files with one fixed mtime (reproducible; the self-test checks it), hashes, signs
the entry through igneum-ota-sign sign-ui with the key in ~/.config/igneum (never read or printed here), copies the
bundle into the folder's ui/ and hands ui.json to publish-manifest.sh --ui, the one writer of the signed manifest,
which verifies the entry and the bundle's hash before signing; --dry-run writes nothing, --verify reads the live
manifest back against dl/<token>/ui and dl/public/ui; --no-ui withdraws the channel. Both self-tests sit on the one
gate. docs/plans/ui-ota.md: the shape, the engine, the security notes, the operator recipe, the tests, per tier.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit e02f5e14d0)
"Interface <version>, over the air, <date>" or "built in", with the help line for pending, refused, held back and
skipped; the "Use the built-in interface" switch (POST /api/ui/builtin). The page posts /api/ui/health once after its
first paint, or the first script error it catches before that; when the served interface changes under it, the page
reloads itself only when idle (no input focused, no sheet or update card open, on the dashboard). ui-ota.test.mjs
covers the words, the reload rule, the ping and that the page loads no remote script.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit b6a0fd0d75)
Seen on igneum-build-1 at 10:39Z on 7 October 2026: both slot files flock-held and empty while two suites ran and
three waiters queued, so the header read 0/2. The holder's missing line is the build-server agent's to fix; the
collector no longer reports a full box as free either way, and the page shows "held, no holder line" in the lane.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The build-server agent has igneum-build-2 and -3 on order (suites and benches; proving and the fast-time nodes). The
installer takes a host argument (N reads ~/.config/igneum/build-server-N, or build@<ip>); the pusher drops the Mac and
PC facts on every box it has a host file for, pulls each box's file and publishes boxes[] (box stays the first for the
old shape); the page draws one server section and one crew card per box, reads every box's Caddy feed in parallel
(build, build-2, build-3.igneum.network) with the edge copy filling any box that does not answer, and merges every
box's builds into the lanes, the timeline and the analytics. Nothing changes for build-1 until the host files exist.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The slot line now ends "; kind=<k> nice=<n> cores=<c>; agent=<a>" (build box scheduling, 69034dc0): lib.mjs lifts the
three into their own fields and keeps the bare label; the old line shape still derives its kind. The collector reads
/srv/builds/_locks/wait-<pid> and gate-pending-<pid> directly (dead pids skipped), marks a gate with priority gate
and sorts it first; the page shows the class line on Now building and Queue cards, a "gate queued" pill, and the class
in the closing lines. JSONL nice, cores, jobs and priority carried through. 14 tests pass.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
(cherry picked from commit 9fce5742a6)
Main's order of 7 October 2026 (igneum-build-2, AX162-1, and igneum-build-3, AX102-1, on order). lib.sh: bs_box_file N and
bs_route <class>; a class whose box has no host file yet falls back to box 1 and says so. run-from-mac.sh --box N <ip> provisions
igneum-build-N and writes build-server-N. tools/build-remote.sh --box N overrides the route; --priority gate always runs on box 1;
the proving crate routes to box 3. README.md: the kind map and the project lead's rule, no mining on any Hetzner box, ever (nodes, builds, tests,
benchmarks and CPU proving only; the pool's fast-time network mines on rented GPU pods, never on build-3). capacity/run.sh refuses a
job that would start igneum-miner mine or a GPU worker, whatever SEQUENCE says.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The slot line now ends "; kind=<k> nice=<n> cores=<c>; agent=<a>" (build box scheduling, 69034dc0): lib.mjs lifts the
three into their own fields and keeps the bare label; the old line shape still derives its kind. The collector reads
/srv/builds/_locks/wait-<pid> and gate-pending-<pid> directly (dead pids skipped), marks a gate with priority gate
and sorts it first; the page shows the class line on Now building and Queue cards, a "gate queued" pill, and the class
in the closing lines. JSONL nice, cores, jobs and priority carried through. 14 tests pass.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The package's design system, layouts and interactions on our tokens and self-hosted fonts (site/site.css section 10),
our facts, numbers, records and live services on every page. The home page: the GPU hero (site/hero-scene.js, continuous
while in view, labelled as drawn), the live DAG panel on scenes/feed.js and live-dag.js 2.0.2, three reasons, the
economics split, the download row with the OS marks, the ledger line. The miner page: the download layout, platform
tabs stamped from the live download index with sha256, the steps, the card table, the fee section (the one source of
/dev-fee), the FAQ. The wallet page in the package's wallet-hero layout, the window slot ready for the wallet redesign.
The secondary pages rendered by the build from one source each: /claims and /randomx from the litepaper's sections,
/dev-fee from the miner's fee section, /provenance from docs/provenance.md through the scrub, /journey from
journey.json; /benchmarks rewrites to /miners, /connect redirects to /metamask. The generated pages (bench, bench
table, ledger) restyled onto the page hero and the docs layout with a section filter. The footer's secondary links point
at the real pages. The copy pass on every served page (docs/plans/site-copy-pass.md). The package's data tests and the
hero loop test in the gate (tools/site-redesign). No sample data is served; the package's review chrome is not.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Main's order of 7 October 2026 after a load of 190 on 96 threads (a release join bench and the 0.3.19 app gate starving each
other, 'builds' of 16 minutes). tools/build-remote.sh resolves a class from the cargo subcommand and --priority: test and bench are
the bounded class (nice 10, the last 32 cores, -j 32) unless --priority gate (nice 0, the full set, the box's own jobs rule);
builds and checks are unchanged. remote-run.sh applies renice and taskset to the command's subshell, caps the jobs, lets a queued
gate (gate-pending-<pid>) take the next slot ahead of suites and benches, and prints nice and cores in the RESULT line and the
JSONL line (nice, cores, priority). The slot label carries '; kind=<k> nice=<n> cores=<c>' before '; agent=', so the dashboard's
job card shows why a job is slow. --plan prints the resolved class without the box; tools/ci/build-kind-default-check.sh (in the
gate) holds the five shapes. Smoke on the box: a suite at jobs=32 nice=10 cores=32, a gate at nice=0 cores=96.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
