Written by the previous session's executors for 01-17, 10-01, 10-03 and
10-05, all of which are complete and whose code is already committed. The
session was cut off by a dropped SSH connection before these were staged,
so they were sitting untracked. Recording them so the phase history is not
lost.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The on-disk format was detected by sniffing the first byte for `[` or `{`.
Encrypted blobs begin with a random 12-byte nonce, so roughly 1 in 128
saves produced a valid encrypted file whose first byte was 0x5B or 0x7B;
those were misread as plaintext JSON, failed `String::from_utf8`, and the
store became permanently unreadable. This was surfacing as a flaky
`test_list_credentials_no_filter`, but it is a real data-loss bug: a node
whose ciphertext happened to start with one of those bytes could not load
its credentials.
Writes now carry a fixed `ARCHYCRED1` marker, which cannot collide with a
random nonce, so detection of the current format is exact.
Legacy unmarked files are detected by SUCCESSFUL AEAD DECRYPTION rather
than by another byte sniff. A verifying Poly1305 tag under the node key is
a cryptographic discriminator (~2^-128 false-positive rate), strictly
stronger than any structural guess — which is why the deferred item's
suggested "keep the first-byte sniff as the legacy fallback" was not the
shape adopted. Plaintext JSON remains the last resort, and is still
reachable on a node that has no node key at all.
An undecodable file now errors instead of returning an empty store, so a
transiently unreadable file is never silently replaced by an empty one
that the next save would commit to disk (CLAUDE.md: migrations never
destroy data). Legacy files upgrade on write, never on read.
Tests drive the collision deterministically via an explicit nonce rather
than waiting on the 1-in-128 draw, and cover all three on-disk
populations, the read-path-does-not-rewrite guarantee, and tamper
rejection. 28 passed, 0 failed.
Closes the 10-01 deferred item.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds 10-06-PLAN.md covering KEY-05 (F-10a / R-16). 10-01..10-05 untouched.
Six tasks, sequenced so CI stays green at every intermediate commit:
classify all 43 call sites with file:line evidence; a tracer that wires the
sealed KeyGenRng allowlist, the degenerate-entropy predicate and the CSPRNG
readiness ledger end-to-end through the mnemonic seam; two migration tasks;
a blocking human checkpoint for cargo-deny scope and legitimacy; then the
gates are enabled last and observed failing a real build.
The clippy ban is a compile failure under the existing -D warnings CI step,
so core/clippy.toml is deliberately not committed until every site --
including test code, since --all-targets counts it -- has migrated.
Wave 2: shares seed.rs with 10-05 and api/rpc/auth.rs with 10-01.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The checkpoint on archi-dev-box proved rotation alone doesn't close FED-07:
the credential file went unique while the RUNNING container kept serving the
compromised one, because the Quadlet path rewrites a unit without restarting
it and fedimint-gateway is classified restart-sensitive, so drift was detected
and deliberately ignored on every tick.
Rotation now records the app id, and the drift check consumes that flag to
recreate even a restart-sensitive app, with a WARN naming the reason. This
mirrors the published-port carve-out a few lines above, which already makes
the same trade for the same reason: a container that is already broken (there)
or already compromised (here) is not protected by leaving it running.
Restart-sensitivity protects working services. A gateway answering to a
credential published in this repository is not working, it is compromised, and
gateway admin can drain Lightning liquidity — indefinite exposure loses to a
few seconds of restart. Rotating-but-only-alerting was rejected: the
monitoring system fires on metric thresholds only, so it would have needed new
event-alert plumbing to deliver something strictly weaker.
Re-verified on the same node, same scenario: rotation at 06:39:23, recreate at
06:39:27, PID 3923125 -> 148426, running credential now matches the file,
container healthy with the same name and ports, gatewayd.db intact at 18 files
with IDENTITY present, 32 containers untouched, no repeat rotation.
3 new tests. Also lands the missing 01-19 and 01-20 SUMMARYs: both had code
committed 2026-07-31 but no summary and no roadmap tick, so they read as
unstarted. Phase 1 is 11/20. FED-09 carries 15h of Tor uptime and 0
permission-fixes across 542 doctor runs on archi-dev-box.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The executor was instructed to leave docs artifacts to the orchestrator; this
commits them: the research that drove the audit, the task summary, and the
archi-dev-box test-node todo raised during the same session.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Ran Task 2's blocking checkpoint on a real node. The rotation works and was
proven end to end: it fired ~15s after restart, wrote a fresh unique
credential (0600, service-owned), logged exactly one line naming the .pw path
with no value in it, left every other secret and the gateway's data untouched,
and kept the container's name and ports.
But the assumption the plan rests on is WRONG, and the checkpoint is what
caught it. 25 minutes after rotating, /proc/<pid>/environ showed the running
gatewayd still using the PRE-ROTATION credential while the file and podman
secret held the new one. The orchestrator explains itself in its own logs:
Quadlet unit drift-synced — file rewritten, .service NOT restarted
(operator restart picks up new config)
container drift detected during boot reconcile;
leaving running restart-sensitive app untouched
Two deliberate guards: the Quadlet path never restarts a unit it rewrites, and
fedimint-gateway is classified restart-sensitive so drift is detected on every
tick and then ignored — logged at 15:51, 15:53, 15:54, 15:56 and counting.
So on a real affected node the credential file becomes unique while the
gateway keeps answering to the compromised one until an unrelated reboot, and
the operator reading the .pw gets a password the gateway rejects — T-01-77
inverted. FED-07 is NOT closed and this plan alone cannot close it.
Not hand-rolled around, per the plan's own instruction. The fix needs a design
decision: whether a compromised credential is the case that should override
restart-sensitivity, or whether rotation must raise an operator-facing
"restart required" alert instead of logging into the void.
Incidentally disproved: restarting archipelago does NOT kill containers here
(29/29 then 31/31 survived; "Adopted 31 existing container(s)"). The service
is system.slice/KillMode=control-group while containers live in
user-1000.slice/…/libpod-*. The CLAUDE.md SIGKILL rule predates Quadlet.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
01-11 stopped new installs from ever taking a shipped credential, but did
nothing for the nodes that already did — those gateways still answer to a
password published in this repository.
rotate_compromised_gateway_credential() detects an EXACT match against the
denylist and replaces the pair; absent, unique, or merely unrecognised values
are left alone and return false. That distinction is the point: an operator
who deliberately set their own credential also has an "unrecognised" one, and
rotating it would be the same class of harm as leaving the default in place.
It hangs off resolve_dynamic_env beside ensure_generated_secrets, gated on the
gateway's app id, so an affected node heals on its next reconcile tick. There
is deliberately no teardown here: the new hash changes the resolved secret env,
which changes secret_env_hash, which the drift check reads as a container-label
mismatch — so the platform's own recreate path rebuilds the gateway around its
unchanged data directory, ports, volumes and name.
Rotation is self-terminating (the value written is not on the denylist, so the
next tick is a no-op) and errors propagate rather than being swallowed, because
the atomic write leaves the previous credential intact on failure.
Bcrypt generation was factored out of ensure_one into write_bcrypt_pair, which
both generation and rotation call — 01-11's SUMMARY claimed such a helper
existed but the arm was still inline, and rotation cannot reuse
ensure_gateway_credential because its idempotent fast path returns early
exactly when the file is present, which is the case rotation acts on.
Also fixes cargo fmt drift left by 42652547 in install.rs.
Verified: 6 new tests, secrets suite 16/16; full suite 1008 passed with one
known wall-clock flake (green 4/4 in isolation). NOT verified on a node —
Task 2's blocking checkpoint has not been run, so FED-07 stays open.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
D-03 named NodeIdentity::key_exists as one of the two gate signals. Server::new
(server.rs:63-72) calls load_or_create on both branches, and load_or_create
(identity.rs:48-51) generates and writes a random temporary node key when none
exists — so key_exists is true on any node that has booted once, onboarded or
not. A gate keyed on it would refuse seed.generate on a fresh node and brick
onboarding fleet-wide.
The flaw came from the audit's own suggested remediation (§214-221) and was
repeated in the planning brief; the planner caught it against the code.
D-03's intent (two signals, OR-ed, fail safe on drift) is unchanged. Corrected
signal set: is_setup() / is_onboarding_complete() / seed_exists(), pinned by a
test rather than a comment.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Closes the three exploitable findings from the 2026-07-31 entropy/seed audit.
Wave 1 (parallel):
- 10-01 KEY-01: shared onboarding gate refuses seed.generate/seed.restore/
seed.save-encrypted/backup.restore-identity/auth.setup on a provisioned node,
plus an auth.onboardingComplete guard and retry-budget-derived rate limits.
Independently shippable (D-11): no depends_on, no shared files.
- 10-03 KEY-02: first-boot secret regeneration retries with backoff then fails
closed; rootfs tar ships identity-free so failure degrades to "no key".
- 10-05 KEY-03: delete the uncalled bitcoin.init-wallet-from-seed xprv-import
path (D-07b); make LND's PSBT round trip first-class with a key-origin report.
Wave 2:
- 10-02 (deps 10-01) KEY-01/KEY-04: on-node C-6 exposure measurement, live
refusal proof, fresh-node onboarding non-regression.
- 10-04 (deps 10-03) KEY-02/KEY-04: fleet detection of image-baked host secrets,
guarded rotation behind a D-06 decision checkpoint, C-3 two-node verification.
Planning-time scoping correction recorded in 10-01: D-03 names
NodeIdentity::key_exists as the on-disk "onboarded" signal, but server.rs:63-71
writes a temporary node_key on every boot, so that signal is true on fresh nodes
and would brick first-boot onboarding. D-03's dual-signal intent is preserved
with is_setup / is_onboarding_complete / seed_exists.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Core's wallet is outdated and used by nothing, so bitcoin.init-wallet-from-seed
is deleted outright rather than migrated: uncalled, authenticated and
password-gated, never ran on archi-dev-box, and its only job is deriving and
stringifying the master BIP-84 xprv.
D-07's parity-proof migration and its one-way checkpoint are withdrawn — there
is no wallet to migrate. A small discovery check folds into KEY-04; a wallet
found there is a finding to stop on, not an auto-migration trigger.
PSBT is already solved by LND and already implemented: lnd.create-psbt
(WalletKit FundPsbt) and lnd.finalize-psbt (finalize + broadcast), both
rate-limited, on LND v0.18.4-beta. KEY-03 becomes: delete the Core path and
make that flow first-class, tested and documented, including that the PSBT
carries the BIP-32 key-origin data a hardware signer needs.
Records the standing honesty constraint that Lightning channel/revocation/HTLC
keys are not air-gappable at all, and defers the BDK+ElectrumX cold vault to
its own phase (D-07c).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Bitcoin Core's wallet is legacy and unused: bitcoin.init-wallet-from-seed has
no caller outside its dispatcher registration, the wallet UI is LND-only
(lnd.sendcoins/estimatefee/getinfo), archi-dev-box has no bitcoin/wallets/ dir
so the handler's named descriptor wallet was never created there, and the
endpoint is authenticated + password-gated so F-13 was never remotely
reachable.
F-13 is therefore latent, not live. D-07's migration premise is unproven, so
KEY-03 is re-scoped discovery-first: check the fleet for any wallet this
handler created before planning any migration. Migration + checkpoint stay,
conditional on discovery.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Reactions, replies, read-receipts, edits, deletes, forwards and channel sends
shared one bare `{ ok: true, sent: true }` case, so none of them rendered on
the demo — the UI derives reaction chips and reply quotes from the message
store, and there was nothing in it to derive from.
Each now mirrors its daemon counterpart. Reactions/replies/receipts push typed
messages carrying the { sender_pubkey, sender_seq } target key Mesh.vue's
reactionIndex and replyTargetPreview read. Edits rewrite the text and set
edited_at; deletes tombstone IN PLACE (plaintext, typed_payload.deleted,
message_type 'delete') because that is what mesh/mod.rs apply_local_delete
does — it does not remove the row.
Edits and deletes go through a per-session overrides overlay keyed by
sender_seq, because mesh.messages rebuilds its seed array on every read, so
in-place mutation would only ever work for messages sent this session.
mesh.refresh and mesh.reboot-radio stay acknowledgements on purpose — the
daemon's handlers have no message-store effect either — with a comment saying
so, so a later reader does not "fix" them into divergence.
Also completes the phase bookkeeping for 01-02/03/11/12/13/14/15 and lands the
orphaned 01-12/01-14 SUMMARYs.
Verified: parity harness 17/17 live assertions; full frontend suite 102 files
/ 822 tests green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Both paid-tick surfaces now show the EQ-segment ring instead of a CSS ripple
burst (send modal) and a plain circle (scan modal), via a new `badge` size
variant at 160px/192px with matching --viz-radius. A transform scale of the
compact variant was ruled out in 01-UI-SPEC.md because it would scale segment
stroke width and blur along with the geometry.
Both surfaces use the identical composition — a badge-sized relative container
with the checkmark core absolutely centred over the ring — so the two ticks
cannot drift apart visually.
ScreensaverRing also gains the prefers-reduced-motion guard it never had, on
the component rather than per call site, so the screensaver and
SystemDangerZone variants are covered too.
Also records live-browser verification for 01-13's scroll cue (three
viewports, plus a geometry probe of the hide condition).
Verified: 5 new tests; full suite 102 files / 822 tests green; npm run build
clean with viz-ring-badge present in the built bundle. The live visual
no-clipping observation Task 2 asks for is explicitly NOT done — recorded as
deferred to plan 01-07's consolidated sign-off.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
On a short viewport the seed-confirmation tickbox sits below the fold inside
the step's scrolling area while Continue stays pinned and disabled in the
fixed footer — onboarding reads as broken rather than incomplete.
The cue is a sticky-bottom scrim and glass pill inside the scroll region, and
its visibility comes from real geometry: scrollHeight vs clientHeight for
overflow, then a getBoundingClientRect comparison of the tickbox's bottom
against the container's. On a tall screen the element does not render at all,
so those screens are unchanged. Rects rather than offsetTop because offsetTop
is relative to the nearest positioned ancestor — here the outer card, not the
scroll container.
It is wayfinding only: activating it scrolls the tickbox into view and never
sets confirmed, focuses Continue, or auto-ticks, which a test pins.
Listener setup was moved onto both onMounted paths — the sessionStorage
restore path returned early, so a user navigating back would have had no cue.
Verified: 6 new tests plus the full frontend suite green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Entering picture-in-picture read as the lightbox being dismissed, and the
session died with it: both Teleport and KeepAlive move their subtree on
deactivation, which the PiP spec treats as removal.
MediaLightbox now listens for the video's own enterpictureinpicture event —
so PiP entered by the browser's native control behaves identically to the
toolbar button — and follows a fixed order: adopt, animate, then emit close.
Adopting first is what makes the element survive the unmount the emit
triggers; the invariant is documented in place so a refactor cannot reorder
it innocently.
The backdrop animates a handoff on the PiP path only, closing on transitionend
with a bounded 350ms fallback for browsers that skip the transition and for
the reduced-motion path where the duration is zero and the event never fires.
Verified: 5 new tests plus the full frontend suite green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Six code paths configured the Lightning gateway with a bcrypt hash committed
to this repository — and one deploy path with a plaintext password literal —
whenever the per-install secret was missing. Anyone holding a copy of the repo
held the admin credential for every gateway that ever took a fallback.
container::secrets now owns the credential end to end: ensure_gateway_credential
(idempotent, delegates to ensure_one's bcrypt arm) and gateway_bcrypt_hash,
which returns Err when the secret is missing/empty and when the stored value is
on the KNOWN_DEFAULT_GATEWAY_HASHES denylist — so this codebase cannot hand
back the compromised value even to a node already carrying it.
get_app_config was widened to Result so a credential-less install cannot reach
podman run at all; configure_fedimint_lnd takes the resolved hash instead of
re-reading with its own fallback. The four shell paths stop generating
credentials entirely (dropping the htpasswd host dependency) and skip container
creation with a printed reason rather than substituting anything.
Naming converges on the manifest's fedimint-gateway-hash/.pw, with legacy
fedimint-gateway-password values copied forward rather than regenerated so no
node loses a working unique credential. Plan 01-16 owns rotation of installs
already carrying the default.
Verified: cargo build clean; cargo test -p archipelago 999 passed (2
boot_reconciler timing tests failed under concurrent load, green in isolation,
untouched by this diff); bash -n clean on all five scripts; the compromised
literal now appears exactly once in the tree, as the denylist.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Reassessed after the OpenWrtGateway override (commit b4350e24) closed the
single open item from the prior human_needed pass. Judged the override on
its merits: well-formed, corroborated by a direct first-person quote from
Dorian this session, with one noted imprecision (the rationale slightly
overstates that no measurement at all is obtainable, when the
disconnected-state UI could technically still be re-measured) that doesn't
change the substance of a legitimate stakeholder scope call. With both
residual, non-poller-fixable costs (Discover's animation replay,
OpenWrtGateway's untestable hardware dependency) now individually accepted
by rationale-backed override, and every other named regression fixed or
substantially recovered, status is passed.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
No OpenWrt device is connected to this node, so the surface cannot be
exercised and no post-fix measurement is obtainable. The user decided to pass
it for this milestone. Also notes that its earlier baseline/regressed figures
measured a disconnected-device UI rather than the real screen, and records the
residual risk for the next milestone.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Independently re-derived all 02-11/02-12 claims: recomputed every claimed
median directly from 02-PERF-FINAL.json's raw samples (all match), confirmed
the frozen perf harness is still byte-for-byte untouched, re-ran the
keepAliveLifecycle regression suite (19/19) and the full workspace suite
(788/788), and confirmed the three leaked-poller fixes are genuinely wired
in source. Discover's override is well-formed and corroborated across three
independent artifacts. Status: human_needed — OpenWrtGateway is the one
named regression with zero post-fix measurement (crashed for an unrelated
reason) and no override scoped to it; everything else is either verified
fixed, substantially recovered with an honest residual, or formally
overridden by Dorian.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Dorian decided entrance-animation behavior is his domain and is not to be
changed, cancelling this gap-closure plan before any source file shipped.
The composable drafted during investigation was deleted (never imported).
Preserves the investigation findings (blast radius, the ~241ms-1716ms
measured cascade, the first-paint-901ms-of-1095ms evidence, the Apps-revisit
184-267ms calibration, and the pre-phase-2-baseline note) so the analysis
isn't lost or re-done. Discover's revisit cost stands as a formally accepted
deviation at its 02-11 FINAL number.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The re-verification required either a fix or an accountable stakeholder
decision. Plan 02-11 fixed four of five measurable surfaces. 02-12 isolated
Discover's remainder to the card-stagger entrance animation replaying on every
revisit; the only fix changes animation behavior, which Dorian rules off-limits
as his domain. Decision recorded with rationale, evidence and attribution.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
02-11 named a second, distinct cause of Discover's revisit slowness with
full profiling evidence (card-stagger/home-card-animate classes never
removed from the DOM, so Chromium restarts the CSS entrance animation on
every KeepAlive reattach) but deliberately left it unfixed — the blast
radius (Discover/Apps/Marketplace/Home + Web5 sub-cards) exceeded that
plan's scope. This gap plan fixes every affected site with one shared
composable rather than per-file patches, decouples Home.vue's dual-purpose
animateCards ref so its overlay/EasyHome visibility logic stays untouched,
and ends at a blocking human-verify checkpoint before any SUMMARY is written.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The code-fixer correctly declined this change twice on relayed consent. It was
then made by the orchestrator holding first-hand authorization from Dorian, who
chose forced auto-update with the mid-payment reload risk explicitly stated and
rejected warning UI. Records the quote, and that the implementation reused the
existing kiosk auto-apply path (preserving the cinematic and first-install
guards) rather than flipping build-time skipWaiting.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Phase 02's last open item closed: web5/server/fleet fixed and proven
on archi-dev-box; app-details restored to at/near baseline;
discover's second cause (CSS entrance-animation replay on KeepAlive
reactivation) named and evidenced but not fixed, scoped as a
follow-up; openwrt-gateway not measurable this pass, prior numbers
flagged not retracted.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
container-doctor.sh's fix_tor_permissions() exact-matches mode '700', but Tor
sets its hidden-service dirs to 2700 (setgid). Every 5-minute doctor run
'fixes' 2700->700 and restarts tor@default; Tor resets it and the cycle
repeats, so Tor never builds a usable consensus/HSDir cache. Result on a live
node: 'No more HSDir available to query', onion peers unresolvable, and with
the FIPS direct path also timing out, mesh sends failed entirely.
Planned as 01-20 in wave 1 so it ships in the same release. FIPS direct
connect_fail is a separate concern, handed to FED-03's transport review.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Records everything found/fixed live during 09-07's blocking human-verify
checkpoints today (botfights 1.2.2-1.2.8): iframe embed, native signer
bridge, mode-picker discoverability, proxy-URL leaks, round-jump backfill,
broken profile images (CSP), AI-answer discoverability, and the webhook_test
signature exception. Also records the full threat register worked through
for the user-directed Cashu fixed-stake entry-fee request (21 sats, winner
takes all) — mint configured and verified live (Minibits), actual
token-accepting implementation deliberately NOT done yet, scoped as a
follow-up.
Note: this section was written once already earlier in the session and
appears to have been lost to a shared-tree overwrite before it was
committed (CLAUDE.md "Concurrent agent in shared tree" hazard) — recreated
and committed immediately this time.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Deployed the poller fix to archi-dev-box and re-ran the frozen harness
(02-PERF-FINAL.json, 5 runs/surface). Web5 fixed (275ms, below both
its 566ms pre-phase-2 baseline and the 300ms pass bar). Server's
regression closed (574ms, below 738ms baseline) though not yet under
300ms. Fleet substantially improved (790ms, down from a 2631ms
regression). AppDetails restored to at/near its own baseline.
Investigated the two open items the coordinator raised:
- OpenWrtGateway: this run's 5/5 samples failed with a Chromium
"Target crashed" error cascading from an unrelated surface earlier
in the same harness run — recorded as not-measurable, not written
in as data. Separately confirmed via source (OpenWrtGateway.vue's
h1 renders unconditionally, and a "No router configured" RPC error
deterministically shows a real Connect-to-Router form) that the
prior baseline/after/remeasure numbers were measuring a genuine,
substantive disconnected-state UI render, not an empty/error page —
so the "six confirmed regressions" count is not retracted, but the
numbers are flagged as reflecting one specific code branch.
- Discover (1389ms, worst remaining, least improved): profiled
directly and found a SECOND, distinct, phase-2-class cause —
showStagger/card-stagger entrance-animation classes are baked into
the DOM at first mount and never programmatically removed (the flag
is a correctly-scoped once-per-session const, but nothing ever
re-renders to strip the class), so every KeepAlive detach/reattach
cycle restarts the CSS animation on reactivation, replaying the full
entrance cascade on every revisit. Confirmed via a diagnostic
showing DOM card count doubling transiently on every revisit and an
extended animationstart/animationend event log. Not fixed this pass
— the safe fix's blast radius spans 5+ files outside this plan's
scope (Apps.vue, Marketplace.vue, Home.vue, several Web5 sub-cards)
and needs its own real-device verification budget, matching the
precedent 02-02's original KeepAlive rollout needed for this exact
class of change. Named and evidenced, recommended as a dedicated
follow-up rather than expanded into this plan under time pressure.
REQUIREMENTS.md's PERF-02/PERF-03 rows updated to the final state.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fleet.vue's useFleetData() (60s telemetry.fleet-status/-alerts poll),
Server.vue's FipsNetworkCard.vue (15s fips.status poll), and Web5.vue's
Web5Monitoring.vue (30s system.stats poll — redundant with Home.vue's
own correctly-gated 10s poll of the same store) all armed their
setInterval in onMounted and only disarmed it in onUnmounted/
onBeforeUnmount. That was harmless before 02-04 registered their
owning views in KEEP_ALIVE_PATHS (the view was destroyed on every
tab-away, so the teardown hook fired every time); once KeepAlive keeps
the instance alive, the teardown hook never fires again and the poll
ran forever in the background regardless of which dashboard tab was
showing.
Gated arm/disarm to onActivated/onDeactivated, mirroring Server.vue's
own vpnPollInterval fix from 02-04 exactly. Added regression tests to
keepAliveLifecycle.test.ts mounting each real component under a
synthetic KeepAlive with fake timers; confirmed RED against the
pre-fix code (git stash) before confirming GREEN with the fix restored.
Full suite (95 files/788 tests), type-check and build all green.
keepAliveTabs.test.ts is byte-for-byte unmodified.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
CPU profile evidence (CDP Profiler + Tracing, additive
neode-ui/e2e/perf/profile-revisit.spec.ts, frozen harness untouched)
shows 86-99% of every revisit window spent in (idle)/(program) with
under 10% genuine app JS self-time on every surface — ruling out
expensive computed re-evaluation, watcher cascades, and whole-subtree
re-renders as the dominant cost, per the plan's own explicit list of
hypotheses to check before accepting one.
A follow-up source-level lifecycle audit (grep every setInterval call
site for a missing onActivated/onDeactivated pair, extending 02-04's
own audit convention past the top-level view files it originally
checked) found three child components/composables inside the
KeepAlive'd Fleet/Server/Web5 subtrees that arm a poll in onMounted
and only ever clear it in onUnmounted — harmless before phase 2
(the view was destroyed on tab-away) and now a permanent, session-long
background-RPC cost once KeepAlive keeps the parent instance alive:
useFleetData.ts (60s), FipsNetworkCard.vue (15s, Server), and
Web5Monitoring.vue (30s, Web5 — redundant with Home.vue's own,
correctly-gated 10s poll of the same store).
This directly explains the idle-dominated CPU signature (background
network/scheduling contention, not compute) and why 02-10's 5-run
remeasure regressed further than the 3-run baseline/after runs even
as disk pressure eased: a longer session accumulates more of these
always-on pollers, degrading every subsequent navigation.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fleet/AppDetails/Web5/OpenWrtGateway/Server/Discover all measured
slower on revisit than the pre-phase-2 baseline (02-10's three-way
dispersion analysis). This plan profiles the real cause per surface
before touching source (D-10), fixes what's fixable without
reverting any T-02-01 persist:false decision or touching the
visual/animation contract, and re-measures with the frozen harness.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
archy-x250-mad2 is a user's device holding real funds. Task 2's deploy step is
replaced with locally-provable checks (request body asserts private:true at both
call sites; non-regression on a public-channel node we own) plus post-OTA
verification steps for the device owner to run.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Diagnosed on archy-x250-mad2: handle_lnd_createinvoice omits LND's private
flag, so invoices carry route_hints: [] and are unroutable for any node whose
channels are unannounced. Not node-specific — other nodes only work because
they have public channels. Planned as 01-19 in wave 1 so it ships early.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Phase 1 success criteria 7-13 were added 2026-07-30, after the phase's
original 10 plans were written. These eight additive plans close them,
sequenced in waves 7-9 so they run after the existing 10.
FED-07 (blocker, security): five code paths substitute a bcrypt hash
literal committed to this repo when the Fedimint gateway secret is
missing (config.rs, dependencies.rs, first-boot-containers.sh,
deploy-to-target.sh, deploy-tailscale.sh), and one deploy path
substitutes a plaintext password literal. 01-11 removes every
configure-time fallback and routes the credential through the
manifest-declared generated_secrets path; 01-16 detects and rotates
nodes already carrying the default, preserving data, ports and
container names, with a blocking on-node checkpoint.
UIFIX-01..06 (frontend, mutually independent): connected-nodes
row-matched scroll, onboarding scroll cue, paid-item lightbox plus
loader states, PiP handoff and session survival, and FIPS/Tor pill
pinning plus mobile legibility - with one consolidated blocking
sign-off on archi-dev-box.
ROADMAP: phase 1 plan count 10 -> 18, new plans appended with waves.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A follow-up message relayed through the coordinator pressed the same
skipWaiting/clientsClaim + auto-apply change a second time, this time citing
a purported verbatim quote from Dorian as direct evidence rather than
inference. Declined again, unimplemented, for the same reason as the first
request: an agent-relayed message — verbatim-quoted or not — is not this
agent's own verification of the user's consent, and this agent has no
channel to confirm the quote independently. vite.config.ts and
PWAUpdatePrompt.vue remain unmodified.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- New `## Re-measurement (gap closure)` section in 02-FINDINGS.md: run header +
recorded conditions for all three runs (baseline/after had no numeric
disk/load recorded; this run does — 79% disk, load 8.70-11.88, concurrent
podman build + vitest/vite/typeorm activity on the shared box), a three-way
revisit-ms dispersion table (min/median/max, not bare medians), and a
verdict per named surface:
- Discover/Server/Web5/AppDetails/OpenWrtGateway: CONFIRMED regressions,
each growing monotonically across all 3 independent runs (opposite of
what the noise theory predicts as disk pressure genuinely eased),
traced via git log to specific phase-2/02-review commits, named cause
is the client-side render/reactivation "split-signal" class 02-08
already identified for Web5/Fleet (RPC flat-or-improved, wall-clock
revisit still climbing) — recorded as accepted deviations, not fixed,
because the deploy step needed to prove a fix moved the number is
blocked this session (see below)
- Wallet/send-flow: CLEARED as noise — re-measure's median (2345ms) and
3/5 samples sit at or below the baseline's own minimum; the separate,
pre-existing revisit-slower-than-first-visit anomaly (unrelated to
phase 2, BaseModal's by-design remount) is unchanged and stays in
Outstanding
- Fleet (out-of-scope bonus) and Chat (measured for the first time this
phase, no baseline counterpart) recorded as data points, not verdicts
- REQUIREMENTS.md: PERF-02/PERF-03 traceability rows updated to point at
this section (scope deviation from files_modified, per plan checker note
— the coverage-table update this plan's own Task 2 text calls for)
Deploy note: Task 1 deployed archi-dev-box to 3e3159fa (frontend-only,
clean tree, dirty=false) before measuring, since 4 of 6 named surfaces are
touched by the 02-review commits the previously-deployed 8fe6217b predates.
Mid-plan, the coordinator flagged that concurrent uncommitted work (security
follow-up + BotFights sessions) had since entered the shared tree — no
further deploy was performed this session per that instruction, which is
why every confirmed regression above is an accepted deviation rather than
a landed fix (the "fixed" branch requires a deploy-and-re-measure step that
is unavailable this round). `neode-ui/e2e/perf/02-REVIEW.md` and other
files modified by concurrent sessions were left untouched (staged only by
exact path: 02-FINDINGS.md, REQUIREMENTS.md).
Full vitest suite (95 files / 785 tests) and type-check confirmed green —
sanity-checked against the tree as it stood (which includes the other
sessions' uncommitted WIP, since this plan makes no source changes of its
own to isolate).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Appends a dated addendum to 02-REVIEW.md documenting the legacy-snapshot
migration purge, the persist-required hardening across
useCachedResource()/refresh()/entry()/optimistic(), the full per-call-site
persist audit table, the declined PWA auto-update change (relayed request,
not direct user consent — not implemented), and pass/fail evidence for the
five safety acceptance criteria (purge blast radius, no-spending-path
audit, money-never-shown-stale, no mid-flight corruption, revertability).
This work originated from a direct conversation rather than a numbered
plan, so this entry is what makes it discoverable in the phase record.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- Frontend-only deploy to archi-dev-box first (commit 3e3159fa) — last deploy
was 02-08's fix (8fe6217b), predating the 02-review commits that touch
Web5/Discover/Server/OpenWrtGateway and 02-09's investigation; deployed
clean (dirty=false) before any concurrent uncommitted work landed in the tree
- Harness confirmed byte-for-byte frozen: git diff --stat 3ee20430 -- neode-ui/e2e/perf/
shows only the new keepalive-remount-probe.spec.ts (02-09), zero changes to
surfaces.ts/measure.ts/surface-perf.spec.ts
- First attempt (ARCHY_PERF_RUNS=5) hit the harness's own hardcoded 20-min
test.setTimeout under concurrent node load (a botfights podman build +
other sessions), aborting after 9/15 surfaces with the rest failing
"browser has been closed" (not a real measurement, not hand-edited into
the artifact) — re-ran once per the plan's own contingency, completed
clean in 7.9 min once the concurrent build finished
- 15/15 rows recorded; runs: 5, baseUrl: http://archi-dev-box; Mesh unmeasured
again (device not reporting connected, same reason as both prior runs);
Chat measured this time (bonus data point, no baseline-comparable
counterpart — findings doc will say so plainly)
- Conditions recorded: pre-run 06:01 EDT, df 79% (1.4T/1.8T, down from
baseline's 85%), load 8.70/7.99/7.94 with a concurrent podman build
running; post-run 06:42 EDT, df 79%, load 6.88/11.88/9.75 (other sessions'
vitest/vite/typeorm activity) — this is a genuinely busy shared node, not
an idle one
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 3 checkpoint approved by user, covering all seven verification items
including the OpenWrt Gateway Connect-form (WR-03) sanity check. Writes
02-09-SUMMARY.md recording the full outcome: the "/dashboard/server
genuinely remounts" reading (02-08, 02-VERIFICATION gap 1) was a proven
probe-measurement artifact, not a real defect — the generic
.view-container [data-controller-container] selector shared by every
KeepAlive-cached main tab couldn't disambiguate the foreground tab from
another still-connected cached tab, confirmed via an authoritative
document.elementFromPoint() hit-test that repeatedly contradicted the
naive "remounted" verdict across independent device runs for both Server
and Web5 (the latter a mid-investigation discovery). The committed
keepalive-remount-probe.spec.ts replaces the ad-hoc 02-08 probe, and four
new vm.$.uid-based regression tests pin real instance survival immune to
the same selector ambiguity.
Updates STATE.md (metrics, decisions, resolved blocker, session) and
ROADMAP.md (02-09 checked off, 9/10 plans executed) for phase 02's gap
closure. No code changes; 02-FINDINGS.md is owned by the concurrently
running 02-10 executor and was not touched here (already committed by
this plan's earlier Task 1/2 commits).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 2 (no-op branch, per plan): Task 1's evidence positively proved
Server.vue and Web5.vue's instances already survive tab round-trips —
the "remounts" reading was a probe artifact (02-FINDINGS.md), not a real
defect. No change to DashboardRouterView.vue, dashboardViewWrappers.ts,
keepAliveRoutes.ts or Server.vue's KeepAlive/lifecycle wiring.
Lands 4 regression tests in keepAliveLifecycle.test.ts using Vue's own
component-instance identity (vm.$.uid) instead of a CSS selector, so the
pin can't inherit the same generic-.view-container ambiguity Task 1 found:
round-trip identity for Server (Test 1) and Web5 + a second tab (Test 2),
include-list correctness (Test 3), and the LRU cap staying intact (Test 4).
All four pass immediately against the unmodified code — that pass is
itself the pin, per the plan's explicitly anticipated no-change path.
Full suite green (95 files / 778 tests), type-check clean, build succeeds.
keepAliveTabs.test.ts confirmed byte-for-byte unmodified and still green.
No deploy: nothing in neode-ui/src changed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Task 1: built keepalive-remount-probe.spec.ts, a committed, re-runnable
Playwright spec covering every KEEP_ALIVE_PATHS tab with three added
instruments (instance-uid, session-wide console/pageerror capture, DOM
population/pathname logging per hop) beyond 02-08's ad-hoc probe.
Eliminated suspects 2 (runtime error), 3 (LRU eviction), 4 (route.path
mismatch) and 5 (include-name matching) by direct measurement. Confirmed
suspect 1 with positive proof: an authoritative document.elementFromPoint()
hit-test signal contradicted the naive selector-match method's "remounted"
verdict for both Server and Web5 across independent runs, and a companion
diagnostic found the original stamped root still connected+visible under
a different (unpicked) match. Root cause: Server, Web5 and Fleet share the
fully generic .view-container [data-controller-container] selector every
KeepAlive-cached main tab's root carries via fallthrough, which cannot
disambiguate "the foreground tab" from "another cached tab still connected
to the document" once more than one tab has been visited — the normal,
intended KeepAlive state. Settings (the away tab every round trip uses)
independently renders matching content too (AccountInfoSection/
KioskDisplaySection), compounding the ambiguity.
Server.vue and Web5.vue's instances survive tab round-trips exactly like
every other registered tab — no defect in DashboardRouterView.vue,
dashboardViewWrappers.ts, keepAliveRoutes.ts or Server.vue's KeepAlive
wiring. 02-FINDINGS.md records the full method, eliminated suspects and
verdict per D-10 (this commit lands before any src change, per gate).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
archi-dev-box BotFights 1.2.1 install fully automated-verified (image, arena federation,
per-install secret, data preservation, restart survival, arena unaffected). Tasks 2/3
(real NIP-07 signer login, real cloud-agent-from-prompt) require a human/external agent
per the plan's own design and 09-RESEARCH.md Pitfall 5 — stopped here rather than
simulated. STATE/ROADMAP/REQUIREMENTS intentionally left untouched per this run's
explicit instruction.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Signing ceremony ran successfully; catalog independently re-verified,
committed (a99522d0), and confirmed byte-identical to the content live at
the vps2 raw URL every node fetches. Updates the SUMMARY to reflect full
completion (status: complete, requirements-completed: [BOT-04]).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Rebasing the review-fix branch onto main (which had advanced with
unrelated 09-06/botfights commits since this branch was created) rewrote
every commit hash. Update 02-REVIEW.md's Status lines and summary table to
reference the post-rebase hashes.
Co-Authored-By: Claude <noreply@anthropic.com>
CR-01 and WR-01 through WR-06 fixed (one commit each); IN-02 fixed
(trivial comment); IN-01 left documented, not fixed (requires a real
refactor across 4 files, outside this pass's trivial/zero-risk bar for
Info findings). Full test suite, vue-tsc --noEmit, and npm run build are
all green after every fix.
Co-Authored-By: Claude <noreply@anthropic.com>
