Editing a prompt file (even externally via "Open in editor") freezes it as a
user customization and blocks future default updates. The Files tab no longer
opens an external editor; it shows each prompt's content read-only in-app via
a new ViewPromptCommand, dropping the file-seeding side effect. Reset to
default and the customized-prompts deviation list are unchanged.
QueuePicker's raw-SQL Queued->Running claim bypasses TaskStateService.StartRunningAsync,
the only other place that broadcasts TaskUpdated on this transition, so the task-list
badge stayed on "Queued" until the run finished. Send TaskUpdated for alreadyClaimed
dispatches in TaskRunner.RunAsync/ContinueAsync to close that gap.
Context-menu "Let Claude handle it" on list rows duplicated the header
button; both routed to the same ListsIslandViewModel command via the
shell, so only the context menu entry (and its now-unused locale key)
is removed. The header button also shared Icon.Broom with Clear Day —
gave it Icon.AgentSuggested instead.
EF Core logs every SQL statement at Information, which BroadcastLogSink
buffers into LogRingBuffer regardless of level, flooding the Log
Visualizer overlay and worker log file with SQL noise that buries
hand-curated business events. Override both to Warning so real
EF Core / ASP.NET Core problems still surface.
Two remaining single-line lists that accumulate entries over time
(table names in Data/CLAUDE.md, HubBroadcaster event names in
Worker/CLAUDE.md) collide when parallel tasks each append a new
entry, same as the AppSettingsEntity/migration-list lines fixed
earlier today. Content unchanged, one bullet per existing item.
update_task, update_task_status, add_task, add_subtask, set_my_day,
abort_merge, review_task, and their batch variants now return a lean
TaskRefDto (id/listId/title/status/sortOrder/isMyDay) instead of the
full TaskDto. Those tools were re-sending the caller's own description
text back on every call, wasting a large share of session context on
list-handler-style runs. get_task/list_tasks/batch_get_tasks are
untouched and still return the full DTO.
## Der Zweifel
Ein autonomer Agent und eine interaktive Session sollen sich unterschiedlich verhalten. Unklar war, ob beide denselben System-Prompt mitbekommen.
## Vorbefund (2026-08-05, beim Verfeinern erhoben — **selbst nachpruefen, nicht uebernehmen**)
Sie teilen ihn **nicht**. Pro Startweg:
| Startweg | Einstieg | System-Prompt |
|---|---|---|
| Autonomer Run | `ClaudeArgsBuilder.cs:69-73`
ClaudeDo-Task: b259a1c9a73846ce833f0b555fb1de79
## Ziel
Ein kleines Icon in der Kopfleiste der Tasks-Insel, das fuer die gerade ausgewaehlte Liste **sofort** eine interaktive ConPTY-Claude-Session im Arbeitsverzeichnis dieser Liste oeffnet — ohne Task, ohne Dialog, ohne Ordnerauswahl.
## Ausgangslage (verifiziert 2026-08-05)
Der komplette Unterbau existiert schon; das ist im Kern Verdrahtung:
- `MissionControlViewModel.OpenAdHocConPtySessio
ClaudeDo-Task: 5d8e6246af9444afa500fbb4546873e7
The usage monitor polled the undocumented OAuth usage endpoint every 60s and
earned 429s. It now polls every 5 min while any task is Running and every
15 min while idle (usage_poll_interval_active_seconds / _idle_seconds, both
clamped to >= 60; the old single usage_poll_interval_seconds key is gone).
A 429 comes back as UsageRateLimitedException carrying Retry-After and adds
exponential backoff on top, capped at 30 min and never shorter than the normal
cadence; the strike count resets on the first success. The schedule arithmetic
is the pure static UsagePollSchedule.NextDelay.
Since the idle cadence is slow on purpose, WorkerHub.RefreshUsage drives
UsageMonitorService.RefreshNowAsync behind a Refresh now button in the Usage
Monitor modal: an out-of-band poll that pushes the loop's next-due time out so
no double poll follows, with a 10s cooldown so click-spam can't earn a 429.
Staleness now measures against the slower (idle) interval so an idle worker
isn't flagged stale just for not polling.
The cache-token task added an ITranscriptUsageReader ctor param to TaskRunner
while the failure-diagnosis task added this test file. Both branches were green
in isolation; the merged result did not compile.
EnsureExists blindly seeded ~/.todo-app/prompts/*.md with the bundled
default and never revisited it, so any file created by opening the
Files settings tab shadowed every later default change permanently
(SuggestImprovement/AskUser sections never reached real runs since
2026-06-04). PromptFiles now hashes what a file was seeded/saved with
in prompts/.defaults.json: Classify() tells missing/current-default/
known-past-default/edited apart, ReconcileStaleDefaults() drops files
that only ever matched a superseded default, and real edits are left
alone and surfaced in the Files tab with a diff + reset-to-default
action. QuarantineOrphans() moves stale-named leftovers (agent.md,
planning.md) into prompts/_orphans instead of silently deleting them.
Wired as a Worker startup sweep (PromptFileRecovery) alongside the
existing OrphanRecovery/AttachmentOrphanRecovery services.
StreamAnalyzer now reads subtype/terminal_reason/errors from the CLI's result
event, and TaskRunner builds a specific ErrorMarkdown from them: max_turns names
the turn budget and points at set_task_config + requeue, api_error passes
through the provider's own message (which carries the reset time), and any
other terminal_reason is appended to the previous generic text instead of
staying invisible. Falls through to the old "exited with code N and no result"
text when there's no terminal_reason at all (a real crash).
## Kontext: Limits sind Fenster, nicht Summen
Die Runs laufen ueber das Claude-Abo. Limits greifen pro 5h-Fenster und pro 7 Tage. Nicht die Wochensumme tut weh, sondern dass ein Agent-Burst ein Fenster leerraeumt, in dem Mika selbst interaktiv arbeiten will.
## Messgrundlage (alle Transcripts unter ~/.claude/projects)
Agent-Runs sind ueber die ganze Historie nur **18,4 %** des Account-Verbrauch
ClaudeDo-Task: 87105f5e-c4f4-4af4-ae60-89cd2e153e3c
task_runs.tokens_in only ever held the API's uncached "input_tokens" field
(off by a factor of ~400,000 on a resumed session), and tokens_out summed
only the last result event instead of the whole session. TaskRunner now
reads each run's cache-read/cache-write/input/output totals from the
session transcript via a new ITranscriptUsageReader.ReadSessionTotalsAsync,
storing the delta against prior runs on the same session so a --resume
doesn't double-count. New task_runs.cache_read_tokens/cache_write_tokens
columns; the Session tab now shows the raw total (what actually counts
against the 5h/7d usage limit) with a breakdown tooltip.
Runaway sessions were the single biggest cost driver: model_presets was
never persisted (stayed code-only), default_max_turns shipped at 100, and
ResolveMaxTurns had no upper bound, so a task/list override could run
hundreds of turns unchecked.
- TaskRunner.ResolveMaxTurns now clamps the resolved value to
AppSettings.MaxTurnsCeiling (new column, default 80) and logs a warning
with task id / requested / effective value when it clamps.
- default_max_turns default lowered from 100 to 40 (entity, EF config,
and the seeded row via the new AddMaxTurnsCeiling migration).
- AppSettingsRepository.GetAsync backfills model_presets with the
shipping defaults on first read instead of leaving the column null.
- Settings > General's per-model preset table and the task/list agent
editor now show a hint when a set max-turns value exceeds the ceiling.
Replace the UniformGrid (all tiles forced to equal size) with a real
Grid + GridSplitters built in code-behind per pane count/column count,
so panes resize individually. The queue strip's fixed 210px column
becomes a drag-resizable Grid column (min 160px) that collapses to 0
when nothing is queued, replacing the old dock-based fixed width.
Adds a "Reading efficiently" section to SystemDefault covering
locate-before-read (Grep/Glob), targeted reads (offset/limit), avoiding
re-reads, and using an exploration subagent as a context firewall for
orientation questions. Based on measurement across 144 real sessions
showing context resend (Read especially) dominates token cost.
All three brief tasks and the nine-child Usage Monitor unit are merged and main
verified green. Correct section 5: "exited with code 1 and no result" is a
catch-all, not a CLI crash — max_turns (sonnet presets default to 30 because
app_settings.model_presets is NULL) and the account session limit both surface
through the same message.
Add a dedicated Usage Monitor & Gate section to Worker CLAUDE.md, wire
UsagePillViewModel/UsageMonitorModalViewModel and the new IWorkerClient
usage surface into Ui CLAUDE.md, and record open verification points
for the pill/modal visual pass and the real gate E2E in docs/open.md.
Data CLAUDE.md already covered the new entity columns and migration.
The pill tests asserted on real localized strings but never set Loc.Current,
which defaults to a key-echo localizer. They only passed because an unrelated
test class happened to install a real Localizer first; adding the usage-monitor
modal tests changed the ordering and the assertions started seeing raw keys.
Initialize Loc.Current in the constructor, matching every other UI test class.
RelayCommand<int> casts its object parameter directly rather than
converting it, so a string CommandParameter ("7"/"30") would throw
InvalidCastException at runtime. Split into two parameterless commands.
Adds a Usage Monitor modal (Worker menu + wired to the footer/Mission-Control
usage pill's Open command): dynamic gauges built from UsageSnapshotDto.Limits
with gate-threshold marks, a stale/blocked-gate band, and Models/Tasks tabs
backed by GetModelUsageAsync/GetTaskUsageAsync over a 7d/30d/custom range.
mika kuns