- Subagents that outlived /clear land in a new session's subagents/ dir with
no meta: borrow the spawning session's agent-<id>.meta.json, else the first
prompt line. Clears the $35.93 '?' row; no launch lacked a description.
- bash_tokens_per_call rose 41% with Bash tokens/day flat (cheap calls fell
28%). Compare bash_tokens_per_day and bash_big_result_share (results >2k)
instead; the Large-Bash lever keys on the share. Tool counters grouped in
ToolPayloads.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TjyHmRrjZssaxXDWcGuCeQ
screen-locker's workout-poke.service takes the phone's POST of a finished
workout on tcp/8773. The flag is persisted like ALLOW_DOT so every re-apply
keeps it, and there is deliberately no wg0 rule: off the home LAN the poke
fails fast and the normal sync credits the workout instead.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018RHwVPyiiQMPh4M3JQwQku
RunnerUp keeps every export on the phone, so each run re-pulled the whole
history into the inbox and re-triggered every inotify watcher there.
pull_from_phone now skips names already in the inbox or in processed/
(processed_dir_for, shared with the importer) and logs one summary line,
escalated to WARNING when any pull failed.
Also: prune evidence/contracts to newest 20 (repo-size cap).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018RHwVPyiiQMPh4M3JQwQku
The daily-limits phone app's launcher icon: charcoal #211D1B field,
hourglass glyph in the shared accent #B8862E. Additive only, so every
other app's icon is byte-identical. A registry invariant now refuses a
new app that reuses another app's glyph.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AcxgxMaQHMSGWzLRpNSkuP
i3blocks' environment has no ~/.local/bin, so `command -v daily-limits`
failed and every left-click on ⏻/🎮 was silently dropped. Call the launcher
by path (DAILY_LIMITS_BIN), detach with setsid, and notify when it is
missing. Also bumps the two pins behind latest stable that turned the
freshness gate red on 433cec27.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QrjUTo1yhgAsUUasdw4bpG
Both blocks read $XDG_RUNTIME_DIR/daily-limits.json (kuhyx/daily-limits,
refreshed every 60 s by a user timer) instead of curling the enforcer and
duplicating the earner table. Stale or missing cache shows ⏱? rather than
old numbers; unknown earners are shown with '?'; left-click opens
daily-limits --gui. Steady-state tick: bash only.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QrjUTo1yhgAsUUasdw4bpG
The shutdown block now shows what can still push shutdown later, like the
gaming-budget block: earners with earned_seconds == 0 from the enforcer's
/api/budget, with a ceiling note once they overflow 23:00. No hint in the
override/SHUTDOWN states or when the server is down.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014qWnrvoCawLZWUpTLJxHue
The issue dashboard now keeps merged-but-open issues (Merge only tags the
opener, kado-no-akari #21) out of `todo` and writes them as `awaiting` in
status.json. The block shows them as "· ⏳ N" between "✓" and "✗", only when
N > 0, without changing the colour (they wait on the opener, not kuhy).
`awaiting` is optional, so the old server's status.json still renders
numbers rather than grey. Still zero forks on the hot path. Prunes the oldest
evidence artifact to stay at the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WXd5qap8DWqZCS9mjh3CzH
[issue_dashboard] shows what the local issue dashboard (~/src/issue-dashboard)
knows, from the status.json its server rewrites on every refresh:
"📋 <to do> · ⚙ <agents working> · ✓ <ready to merge>", green when something
is ready to merge, red with "· ✗ <n>" when an agent or a proof failed, grey
"📋 dashboard" when the file is missing or stale (server not running). Zero
forks on the hot path; a click opens issue-dashboard-desktop, which reuses or
starts the server and opens its window. Behaviour + fork-count test in
tests/test_i3blocks_issue_dashboard.sh. Prunes the oldest evidence artifact
to stay at the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WXd5qap8DWqZCS9mjh3CzH
[book_guard] shows this month's paper-book pace from the state.json
book-guard publishes (green on pace, yellow behind but unlocked, red when
locked or the timer is disarmed, orange when missing/stale); a click opens
book-guard-desktop -- the same way [screen_locker] opens its UI. Behaviour
test in tests/test_i3blocks_book_guard.sh.
meta/requirements.txt: gatelock 0.9.0 (the shared morning-session reader),
plus today's newest stable for anthropic, cryptography, flake8-bugbear,
geopandas, howlongtobeatpy, pyflakes, selenium, mypy, pylint and ruff.
Prunes the oldest evidence artifact to stay at the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FBbkN4ZB4aUPJPzR2prRan
The package now lives in its own private repo at ~/src/wsg-grabber, with
the 14 commits that touched it. /usr/local/bin/wsg-grabber was regenerated
by that repo's install.sh, so nothing here runs it any more.
hypothesis 6.168.2 -> 6.168.3: an upstream release turned the Pre-commit
checks and dependency freshness jobs red on 33f0fee3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F8G3BNETGmbr6aM5cc18g4
Kept videos land in ~/data/cloud/Media/<year>/wsg (~/Downloads without the
cloud folder); trash/ keeps only the newest 200 passes and marks the rest
purged. Legacy keep/ is migrated once. Undo follows a keep through every
place it can end up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01F8G3BNETGmbr6aM5cc18g4
keyboard_is_up read the phone-wide mInputShown, but Android draws the IME
on display 0 even when its target window is on a virtual display, so it can
never cover a VD tap. With two sessions on one phone, the other session's
keyboard on the real screen made dismiss_keyboard press BACK on this
display, popping the route under test and then closing the app.
Prunes the oldest evidence artifact to stay under the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KeVXsWTuwjmZK5uvbw6vqF
setup_homeassistant.sh runs a digest-pinned Home Assistant container with
host networking and installs hass-xiaomi-miot v1.1.5, so the Viomi V3
(viomi.vacuum.v13) is controllable over the LAN from the HA companion app.
setup_wireguard_ssh.sh gains a persisted ALLOW_HOMEASSISTANT flag that opens
tcp/8123 to the LAN subnet and wg0 only. Prunes the oldest evidence file to
stay under the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CGnk48tUeJteJWEQ7UhKDd
A no-cache run (all pre-commit caches deleted, workflow_dispatch
36301788851) exercised the fallback for the first time: the first
build failed on sdl-config in 7 s, the headers installed in 21 s, the
retry built the env, and the job went green -- but the expected
traceback tripped setup-python's problem matcher and left an error
annotation on the green run. The first attempt now logs to a file; the
retry still prints in full.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KFwA9nymhUdTrVF65kigTV
Measured on c73cfd3: cold 59 s / warm 54 s / 47 s -- inside 60 s with
little margin. shellcheck is a language: system hook with no env, yet
its group restored (and on a miss saved) the ~120 MB hook-env cache.
The prettier/shfmt version-record step moves from rest (the slowest
group on the warm run, 49 s) to pylint (the fastest, 31-37 s).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KFwA9nymhUdTrVF65kigTV
The first run of the split workflow restored the pylint hook env from
the old monolithic cache, yet still spent 30 s on apt for pygame's SDL
headers, because the step was gated on an exact cache-key hit. That made
the pylint group 71 s. Now the env build is attempted first (pre-commit
installs the env, then has no files to lint) and the headers are
installed and the build retried only when it actually fails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KFwA9nymhUdTrVF65kigTV
Target: every push workflow <= 60 s. No check removed or weakened.
- Pre-commit checks (164 s): mypy, pylint and shellcheck each get a
matrix job, everything else runs in `rest` with SKIP=<those three>
(pytest-coverage stays skipped everywhere, as before). The full
requirements.txt install (49 s, plus the SDL headers for a pygame
source build) is replaced by `pre-commit pyyaml packaging` -- the only
third-party imports of the language: system hooks. Per-group
pre-commit cache with same-interpreter restore-keys; SDL headers only
on a pylint env rebuild; zsh only in rest; shellcheck only in its group.
- Python tests (87 s): setup-uv with a cached wheel store instead of pip,
pytest -n auto -q -p no:sugar (2051 tests, 100.00% under xdist locally
across three seeds), apt install tried before `apt-get update`.
- Shell tests (63 s): the lib/tests runner loop moves to its own job so
it overlaps with bats and the side-effect-free tests; runners still run
sequentially within it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KFwA9nymhUdTrVF65kigTV
The dependency-freshness gate went red on hypothesis 6.168.2, released
2026-09-27. Newest stable, exact pin, nothing else changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KFwA9nymhUdTrVF65kigTV
New [gaming_budget] block reads steam-backlog-enforcer's /api/budget every
30s: "🎮 2h00 left · 6h00/8h00 +📖1h". Colours follow the enforcer warn_at
(1h/30m/10m), blocked is red, a dead server shows the cached value as stale.
Unearned bonus sizes come from the enforcer config via jq --slurpfile.
Prunes the oldest evidence artifact to stay under the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CD7nSVE2dKvByG8hdSi3gm
Android runs one instrumentation per package: a second `am instrument`
force-stops the first ("stop com.kuhy.a11ydump due to start instr"), and
so does `install -r`. The live two-session run on the Pixel caught two
sessions' dumps killing each other ("Process crashed"). dump_display now
runs install + instrument under a per-serial flock; a dump takes well
under a second, so queueing costs nothing noticeable.
After the fix the same 300 s run had 0 failures across 659 actions, and
the real screen and both apps stayed where they were.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WpA6wW2DKiovQADTdjWJoH
A raw `adb shell input tap` took no lease, so one session's tap landed in
another session's app (2026-09-25). python_pkg.phone_guard classifies every
adb call in a Bash command and, as the Claude Code PreToolUse hook, takes
the matching phone_lease (one app / real screen / whole phone) or exits 2:
key/text input without -d, input on another session's virtual display, or
a lease still held elsewhere after 40 s.
android_ui gains --display self|<id>: it drives only this session's
phone_vd display under that app's lease, and every input carries -d.
uiautomator sees display 0 only, so virtual-display trees come from a tiny
instrumentation APK (A11yDump.java) built from source without Gradle.
Prunes the oldest evidence artifact to stay under the 60-file cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WpA6wW2DKiovQADTdjWJoH
Several sessions can now share one phone: a lease covers one package (the
real screen stays the default scope, so existing callers keep their
meaning), and only phone-wide actions take `--whole-phone`, which is capped
at 300 s and runs its `--cleanup` commands exactly once on release or, via a
detached reaper, on expiry. The serial is always resolved, never the
"default" placeholder that let android_ui and phone_deploy lease different
keys for the same phone (2026-09-25). Adds status --json, holds and whoami
for phone_vd.sh and the raw-adb hook.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012Ajzz98xksbfak8CaunZWC
Filled scalloped canopy over a stroked J-handle (filled because a stroked
dome with ribs reads as a jellyfish at 48dp). AppIcon moves to _app_icon.py,
re-exported from apps.py, so the registry stays under the 250-line cap.
Prunes the two oldest evidence artifacts (60-file cap; still in history).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012Ajzz98xksbfak8CaunZWC
Two sessions driving the same phone interleave their taps: on 2026-09-20 a
tap meant for the workout sandbox opened another session's manga reader,
twice. The lease is a flock-guarded JSON file per serial under
~/.cache/phone-lease/, owned by the nearest ancestor `claude` process (one per
session), refreshed on every android_ui call and expiring 180 s after the
last one, so a quiet session frees the phone without anyone remembering to.
A foreign fresh lease makes a call wait up to 60 s, then exit 3 naming the
holder and pointing at the alternatives (the phone's virtual display, the
headless emulator) instead of inviting a retry loop. phone_deploy.sh takes
the same lease (exit 40).
The oldest evidence artifact is pruned to stay under the 60-file cap; it
remains in history.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01YbBZqUaNWyhHYfzz6zGC7q
