Surfaces prompt-cache hit rate on the Insights page so cache-efficiency
(a key cost-optimisation signal) is visible at a glance.
Backend (api/routes.py, _handle_insights):
- Aggregate cache_read_tokens per model and per day, plus a total.
- Graceful fallback for older agent state.db schemas lacking the
cache_read_tokens column (COALESCE + OperationalError second query).
- New response fields: total_cache_read_tokens, total_cache_hit_percent,
per-model cache_read_tokens + cache_hit_percent, per-day cache_read_tokens.
Frontend:
- Models table gains a Cache column (panels.js + style.css 6-col grid).
- Daily bar tooltip includes the bounded cache hit rate.
- New i18n keys insights_model_cache / insights_cache_hit in all 13 locales.
Correctness fix (maintainer-flagged on #3912): the original computed
hit-rate as cache_read / input_tokens (reads-over-misses), which EXCEEDS
100% on cache-heavy sessions. This uses the FULL prompt total as the
denominator -- cache_read / (input_tokens + cache_read) -- via the shared
api.usage.prompt_cache_hit_percent helper, so it is bounded 0-100% and
means "% of the prompt served from cache". Returns None (rendered as a
muted dash) when there are no cache reads.
Tests (tests/test_insights.py):
- Updated existing assertions for the new response shape.
- Added a non-vacuous <=100% boundary test: a cache-heavy case
(cache_read=9900, input=100) where the OLD formula reports 9900% but the
bounded denominator yields 99%. Verified the test FAILS under the old
formula and PASSES under the fix.
Salvage note: PR #3912's diff was ~99 releases stale (mostly re-derived old
code, illusory +5405). This ports ONLY the real ~190-line feature fresh
onto current master.
Closes#3911
Co-authored-by: TomBanksAU <TomBanksAU@users.noreply.github.com>
Lets a trusted local extension contribute a TTS engine that appears in the
Settings -> TTS Engine dropdown and is used by BOTH playback paths (voice-mode
auto-read in boot.js _speakResponse + the per-message Listen button in ui.js
speakMessage). Mirrors window.registerHermesSkin.
window.registerHermesTtsEngine({ id, label, synthesize(text, opts) })
-> Promise<ArrayBuffer|Blob> of audio; core coerces + plays through the same
<audio> lifecycle as the Edge engine (stop/rearm in voice mode).
Core owns selection, the dropdown <option> (label via textContent, never HTML),
and playback. Built-in ids (browser/edge/elevenlabs) are reserved; invalid
descriptors are rejected; empty/failed synth degrades gracefully. The settings
panel re-adds registered options on render (the markup hardcodes only built-ins).
Unblocks TTS-engine extensions (e.g. a local VOICEVOX engine). Docs in
docs/EXTENSIONS.md; CHANGELOG [Unreleased]. New regression test drives the real
registry via a Node harness (register/validation/reserved-guard/synth coercion).
Full suite: 10893 passed, 0 failed. Verified end-to-end in a live browser with a
companion VOICEVOX extension: engine registers, appears in the dropdown
(browser/edge/elevenlabs/voicevox), and the Listen button routes through the
registered synthesize (both VOICEVOX steps called).
Adds a "Hermes Dashboard" chip to the Settings > Appearance "Sidebar tabs"
grid, bidirectionally synced with the existing dashboard-mode dropdown
(never/always/auto). Chip off -> dropdown "never"; chip on -> restore last
non-never mode; dropdown change -> chip re-render.
The chip is a proxy for the dashboard config via saveDashboardSettings ->
POST /api/dashboard/config; it NEVER mutates hidden_tabs/tab_order (separate
toggle fn from _toggleTabVisibilityChip). loadDashboardSettings guards a stale
config GET from clobbering a newer save (_dashboardSettingsLoadSeq /
_dashboardSettingsWriteSeq), and a failed chip save rolls the dropdown/chip
back. Chip label via textContent (XSS-safe).
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
The streaming path already honored a max_tokens cap from the active profile's
config.yaml, but there was no UI to manage it. Settings > Preferences now has a
"Max output tokens" field that writes a root-level max_tokens override through
the profile config (not settings.json, which the streaming worker never reads
for this), preserving unrelated YAML keys under _cfg_lock. Blank clears the
override so agent.max_tokens fallback resumes; get_max_tokens_status() surfaces
root / effective / fallback so the field shows what a new turn would use.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Renders generic registry post_install guidance + lifecycle requirements in the
Settings extensions gallery card (manifest-driven, no vendor special-casing).
Fixes#4959. Rebased onto current master (squash-applied #4964's net diff so it
coexists with the #4979 sidecar-runtime-status feature that landed in v0.51.697).
Co-authored-by: franksong2702 <franksong2702@users.noreply.github.com>
Codex gate finding on #4979: loadExtensionsPanel({preserveExisting:true}) on the
diagnostics tab treated the 'Loading extension diagnostics…' placeholder as real
content, so a failed refresh returned early (preserveExisting && innerHTML.trim())
and left the panel stuck on 'Loading…' instead of rendering the error. Exclude the
.extensions-loading/.extensions-error placeholders from the preserve check + add a
regression assertion.
Co-authored-by: franksong2702 <franksong2702@users.noreply.github.com>
Keep the Diagnostics tab content visible while status refreshes on repeat visits, and reject malformed or far-future sidecar runtime timestamps before rendering relative age text.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Cron run output is never authored Markdown -- it is logs/JSON/plain text.
Using renderMd() with any heuristic or synthetic fence is fragile:
- Content starting with #, |, > becomes headings/tables/blockquotes
- Embedded triple-backtick in content breaks the synthetic fence
Fix: use DOM-created pre/code with textContent throughout:
- No heuristic, no synthetic fence -- every byte renders verbatim
- Both render paths (initial load + View full output) use the same approach
When viewing cron job run logs (prompt/response), the raw content
was passed through renderMd() which broke formatting for JSON and
plain text (single newlines collapsed, special chars interpreted).
Now detects if content looks like markdown and wraps raw content
in triple backticks so it renders as a proper code block with
preserved whitespace and formatting.
Gate (Codex) caught two SILENT regressions in the original fix: explicitly
hiding the header in the 'else'/'empty' branches also hid populated read views.
- _setMemoryHeaderButtons('read') now shows the header for ALL read views
(incl. read-only Project Context / External Notes), gating only the Edit btn.
- _renderProfileConceptHelp() now uses a new 'help' header mode (shows header,
hides all action buttons) instead of 'empty' (which hid the help title).
Adds tests/test_issue4891_detail_header_visibility.py (proven non-vacuous).
Co-authored-by: luandnh <luandnh@users.noreply.github.com>
The main-view header (and its action buttons) used a CSS :has(.main-view-title:empty)
rule to stay hidden until the title populated; on some mobile browsers / PWA
WebViews that pseudo-class doesn't re-evaluate after textContent is set
dynamically, leaving the header permanently hidden. Explicitly set header
display in the read/create/edit/closed header-button setters (tasks, skills,
memory, workspaces) instead of relying on :has() reactivity.
Co-authored-by: luandnh <luandnh@users.noreply.github.com>
- Add KANBAN_UNASSIGNED_LANE constant and _kanbanLaneKey/_kanbanLaneLabel
helpers so unassigned tasks map to a stable internal key '__unassigned__'
rather than a locale-dependent translated string
- _kanbanLaneNames now places 'default' first, other assignees alphabetically,
and the Unassigned lane last
- _kanbanRenderProfileLanes filters by _kanbanLaneKey, renders label via
_kanbanLaneLabel, and adds kanban-profile-lane-unassigned CSS class
- _kanbanTaskMeta always emits the unassigned label so sidebar shows it too
- _kanbanRenderBoard checks unfiltered total and shows a 'Tasks hidden by
filters' state with a Clear filters button when text search hides all tasks
- Add kanban_tasks_hidden_by_filters and kanban_clear_filters i18n keys to
all 13 locale blocks
- Add .kanban-profile-lane-unassigned header border using warning token
Fixes: on Dynasty board mobile, Ready=3 CLI but default lane showed Ready=0
because those tasks were unassigned and had no visible lane.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When a cron job's model/provider override was set via the form, the dropdown
option value could contain a @provider: prefix (e.g. @custom:9router:chat)
from _apply_provider_prefix in the API response. Since cron jobs store model
and provider as separate fields, the model should be just the bare model name
(e.g. chat) without the provider prefix.
- Add _cronModelBareName() helper to strip @provider: prefix
- Apply in saveCronForm for create, update, and duplicate paths
- Update CHANGELOG.md
The JS module default was false (consolidated) while the server defaults
lane_by_profile=True (lanes). If the board painted before _kanbanApplyConfigDefaults
ran, there'd be a one-frame consolidated flicker before snapping to lanes. Matching
the client default to the server default removes the flicker. Co-authored-by: stefan-wisbric
Took master's i18n.js as base (it has all 13 locale blocks + had reorganized the
kanban keys, so the stale-branch conflict block would have duplicated them in
zh-Hant), then inserted #2814's 5 kanban_view_* keys into every locale block after
kanban_new_task — real translations for it/ja/zh/zh-Hant/fr (from the PR), English
fallback elsewhere. All 13 locales now have exactly 1 of each key; locale-parity +
kanban tests green (291 passed).
test_smooth_text_fade.py::test_preferences_ui_exposes_and_saves_fade_text_effect
extracts loadSettingsPanel() via a brace-matcher that does NOT skip comments. The
new structured-code wiring added a comment containing 'isn't' + ''auto' mode' —
the apostrophes were mis-read as string delimiters, scrambling the brace depth so
the matcher closed the function early (before the fade-checkbox wiring), failing
the test. Reworded to 'is not' / 'auto mode' (zero behavior change; node -c clean).
The fade feature + the structured-code feature are both unchanged. Co-authored-by: metaember
The JSON/YAML tree viewer (#484) hardcoded its default: valid json/yaml
fenced blocks opened in Tree view at 10+ lines (`const showTree=lineCount>=10;`).
Make that default user-configurable while preserving current behavior.
- New settings (persisted via /api/settings -> settings.json):
- structured_code_default_view: auto | on | off (default auto)
- structured_code_auto_tree_lines: int 1..1000 (default 10)
- ui.js: replace the hardcode with a sanitized, unit-testable decision
helper _structuredCodeShowTree(mode, threshold, lineCount); Tree/Raw
toggle, Raw <pre><code>, parse-failure fallback, and YAML lazy-load
are unchanged.
- Appearance settings: 'JSON/YAML code blocks' select + 'Auto threshold
lines' input (disabled outside auto mode); wired through the existing
appearance-autosave and explicit-save paths; live re-render on change.
- boot.js seeds window globals; i18n labels added to all 13 locales.
- Default (auto + 10) reproduces the original >=10 behavior exactly.
The profile-switch skeleton (#4671) always rendered a fixed content shape; for
a brand-new/empty profile that flashed a misleading 'loading conversations'
skeleton that resolved to an empty list. Track each profile's last-seen session
count (localStorage, per profile name, recorded after each non-all-profiles
render) and, when switching into a profile KNOWN to have zero sessions, render a
quiet empty-state placeholder instead of the content skeleton. Unknown profiles
keep the content skeleton (safe default); skipped under an active project/source
filter (the count is an unfiltered total). prefers-reduced-motion + light/dark
aware. Frontend-only. 7 source-structural tests; Phase-1 skeleton tests still
green.
Opus ship-gate non-blocking observations (no MUST-FIX), applied to lower future-maintainer
risk on the crown-jewel profile-switch surface:
- clarified the belt-and-suspenders _sessionListSkeletonActive=false at the real-paint site
- added a FOUR-GUARD CONTRACT doc-comment at the top of switchToProfile naming the four
interacting state pieces (embargo / invalidation / skeleton-flag / _wsTreeGen + _switchGen)
so a future edit touching one without the others doesn't silently reopen a window.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Two more edge cases from re-gate 4:
(1) SILENT: a mid-switch /api/sessions that FAILS bypassed the embargo (the drop-check
was only on the success path), so its catch cleared the skeleton + rendered old-profile
cache. Added 'if(_profileSwitchListEmbargo) return;' after the gen guard in the catch too.
(2) CORE: the embargo was set BEFORE the try, so a throw in the synchronous setup could
leak it (the guarded finally only covered the try). Moved the invalidate/embargo/skeleton/
ws-gen setup INSIDE the try so the _switchGen-guarded finally always lifts the embargo.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Codex re-gate 3 found a third clobber variant: point-in-time invalidation isn't enough
because a renderSessionList() can START after the skeleton shows but BEFORE
/api/profile/switch returns. Since the new-profile cookie is only set by the switch
response, that GET fetches the OLD profile's rows, passes the (now-current) generation
guard, and clobbers the skeleton.
Fix: a profile-switch session-list EMBARGO. _setProfileSwitchListEmbargo(true) before
showSessionListSkeleton(); _runRenderSessionListRefresh drops ALL payloads (after the
gen guard, before apply) while embargoed; the switch lifts it immediately before its own
authoritative renderSessionList() (both branches), in the failure-restore, and in a
_switchGen-guarded finally as a safety net so it can't freeze the sidebar on an
early-return/throw. + regression test.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Codex re-gate 2 found a remaining CORE clobber variant: a renderSessionList() already
in flight BEFORE the switch (background poll fetching the OLD profile's /api/sessions)
would resolve after the skeleton appears but before the switch's own renderSessionList()
bumped _renderSessionListGen — so it passed the generation guard, cleared the skeleton
flag in _applySessionListPayload, and painted stale rows.
Fix: _invalidateSessionListRenders() bumps _renderSessionListGen + clears
_pendingSessionListPayload / _renderSessionListQueuedRequest, called at switch start
BEFORE showSessionListSkeleton(). Any outstanding pre-switch response is now stale
(its captured _gen < current) and discarded by the existing guard; deferred/queued
applies are dropped. + regression test asserting the helper exists and runs before the
skeleton.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Reworked from the Codex re-gate (rebased onto fresh master so #4730's gateway-poll
work is preserved, not reverted). Two CORE races + a test-window break:
(1) Skeleton-clobber (sessions.js): renderSessionListFromCache() now bails while
_sessionListSkeletonActive — _allSessions still holds the PREVIOUS profile until
/api/sessions resolves, so a gateway-poll/SSE/timer/cache render mid-switch would
repaint the wrong profile's rows over the skeleton. _applySessionListPayload clears
the flag right before painting (fresh data in hand; superseded responses already
discarded by the generation guard); the fetch-failure path and switch-failure catch
clear it too so it can't strand. NO signature change (kept renderSessionListFromCache()
and _applySessionListPayload(sessData,projData) so the ~10 static tests that anchor on
those markers stay green).
(2) Workspace wrong-profile tree (workspace.js + panels.js): an empty-session switch
reuses the same session_id, so loadDir()'s id guard can't reject a stale pre-switch
/api/list. Added _wsTreeGen generation token + bumpWorkspaceTreeGen(), bumped
UNCONDITIONALLY at switch start (even panel-closed, since loadDir('.') still runs),
captured in loadDir() and re-checked after both awaited /api/list points.
(3) test_issue1116 proximity window 5000->6500 (applyBotName now ~char 5618, still fires).
+ regression tests: skeleton-flag bail/clear lifecycle, ws generation token, failure
restore clears flag; updated #1785 stale-session guard test for the compound condition.
Co-authored-by: rodboev <rodboev@users.noreply.github.com>
Addresses @rodboev + greptile review of #4671:
- panels.js switchToProfile(): pass timeoutToast:false to the /api/profile/switch
POST so a superseded or transient-but-successful switch no longer pops api()'s
generic 'Request timed out' error toast; the generation-guarded catch handler is
now the single source of truth for switch failure.
- panels.js: re-check _switchGen after the awaited renderSessionList() in the
in-progress-session branch so a rapid second switch can't have its workspace
skeleton cleared / a stale toast popped by a slower earlier switch (mirrors the
no-messages branch guard).
- style.css: group labels now fade in via skeletonFadeInDim (ends at opacity:.5)
instead of the shared skeletonFadeIn (ends at opacity:1 with fill-mode:both),
which was overriding their intended resting 50% opacity.
- 4 new static-source regression assertions locking the three fixes; CHANGELOG
Unreleased entry updated (fixed-shape skeleton note + the three review fixes).
The success path clears a stranded workspace-tree skeleton when the new
profile has no bound workspace (both switch branches), but the catch block
only restored the tree `if (S.session && S.session.workspace)` — with no
else. So a profile-switch failure while the workspace panel is open and the
(still-current) previous session has no workspace left the up-front skeleton
shimmering forever: the same skeleton-strand class the Codex/Opus gates
targeted, in the one path they didn't cover. Mirror the success-path
no-workspace handling in the failure path. Adds a static regression
assertion that the catch block clears the workspace skeleton.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
