Replaces the fragile DRAWER_GESTURE_SCREENS global set + defensive
LaunchedEffect(snapTo(Closed)) + closeDrawer callback workaround with a
single source of truth: each screen declares its drawer-gesture preference
via HermesScaffold's new drawerGesturesEnabled parameter (default true).
Architecture:
- DrawerGestureController holds the active gesture state. HermesScaffold
reconciles each screen's preference into it via SideEffect; when a screen
opts out while the drawer is open, the controller closes the drawer itself.
- LocalDrawerGestureController CompositionLocal flows the controller from
MainNavigation down to ModalNavigationDrawer, which reads .enabled and
passes it to Material's gesturesEnabled parameter.
- DisableDrawerGestures() helper for entry screens that don't use
HermesScaffold (Landing, AuthLogin, PairingCodeEntry).
- All 6 Settings drill-down sub-pages pass drawerGesturesEnabled = false.
- NavigationController.closeDrawer removed; navigateTo no longer needs a
synchronous drawer-dismiss hook.
Removes: DRAWER_GESTURE_SCREENS set, both LaunchedEffect(gesturesEnabled)
and LaunchedEffect(drawerGesturesEnabled) blocks, drawerGesturesEnabled
local state, closeDrawer callback on NavigationController.
Fixes the recurring drawer-scrim stuck-open bug class (PRs #445, #454, #455).
Verified: ktlintCheck, assembleDebug, testDebugUnitTest all green locally.
Co-authored-by: Hy4ri <hy4ri@users.noreply.github.com>
Split the 2,267-line ChatScreen.kt into 11 single-purpose files under
ui/chat/components/, keeping ChatScreen.kt as a thin (~530 line)
entry compositor that delegates rendering to the new components.
Files extracted:
- ChatConnectionBanner.kt — top banner (reconnecting/disconnected/auth-expired)
- ChatMessageList.kt — message list + search bar overlay + empty state
- ChatComposer.kt — input bar + attachment chips + slash dropdown
- ChatSearchBar.kt — CompactSearchInput + SearchBarRow
- ChatDialogs.kt — ClarifyDialog, SudoPromptDialog, SecretPromptDialog, ReloginDialog
- ChatIndicators.kt — ThinkingIndicator, ReasoningIndicator, SubagentIndicatorRow
- ChatScrollFab.kt — scrollToBottom utils + FAB
- ChatLifecycleEffects.kt — session switch, lifecycle observer, auto-scroll, error snackbars
- ChatReactions.kt — ReactionHeartsOverlay + FloatingHeart animation
- ChatStreaming.kt — StreamingBubbleWithTypingEffect
- ChatLoadingOverlay.kt — full-screen semi-transparent loading overlay
No behavioral change. All composables are public where they need to be
called from ChatScreen, otherwise keep their original visibility.
Unblocks #583 and #584 which were previously too risky to attempt in
the 2.2k-line file.
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Add ToastHost interface + override keyword to the 3 ViewModels that
declared clearToast() without implementing the interface. This aligns
them with the other 17 ViewModels and enables polymorphic substitution
where ToastHost is expected.
Fixes#614
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat: auto-scroll long chat title once on load + on tap
* fix: resolve state delegate + suspension compile errors in AutoScrollingTitleText
* fix: use Int scroll offset for animateScrollTo
* fix: measure real container width for title marquee (was always Infinity)
* fix: ease the title marquee scroll (FastOutSlowInEasing + glided return)
* fix: address agy review — cancel overlapping scrolls + one-shot on-load guard + clickable on box
* fix(ui): smooth out top app bar collapse transition
Use dynamic layout padding based on scroll offset to prevent blank gap at the top of content when app bar collapses.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat(#606): support auth mode + bearer token when adding HTTP MCP servers
- Extend AddMcpServerRequest with auth (none|header|oauth) and bearerToken
- Add auth mode FilterChip selector in HTTP add-server dialog
- Conditionally show masked bearer token field when auth == header
- Wire new fields through McpServersViewModel request builder
Backend contract: exactly one of url/command required; bearer_token
requires auth == "header"; args/env only for stdio. SnakeCase naming
strategy auto-serializes bearerToken -> bearer_token.
Fixes#606
* fix(#606): address agy review — remember PasswordVisualTransformation + FlowRow for auth chips
- Wrap PasswordVisualTransformation in remember {} to avoid per-recomposition allocation
- Use FlowRow for the 3 auth-mode FilterChips so they wrap on narrow screens instead of overflowing
* refactor(settings): drill-down navigation replacing tabbed layout
Replaces the 4-tab settings layout with a native Android-style drill-down:
a single category list (SettingsScreen) where each row pushes a dedicated
sub-page onto the Navigation3 back stack (SettingsConnection / Appearance /
Chat / NavBar / Behavior / About). Each sub-page is its own NavKey destination
registered in Navigation.kt and reuses the existing section composables.
- SettingsScreen becomes the category list with icon + label + summary
(Appearance row shows live theme + language, e.g. "Dark · 한국어").
- New SettingsSubPages.kt hosts the 6 full-screen sub-pages.
- AboutSection promoted out of SettingsScreen into its own sub-page.
- SettingsViewModel remains the single source of truth; no persistence changes.
- Adds settings_summary_* strings for list subtitles.
Consumes the in-app language switcher from the prior commit (AppearancePage
passes appLanguage through to AppearanceSection).
* [bug] fix settings sub-page back button opening drawer on first tap
The ModalNavigationDrawer wraps the whole app and stays gesture-enabled on
the Settings root (a DRAWER_GESTURE_SCREEN). Drilling into a sub-page
(SettingsConnection/Appearance/Chat/NavBar/Behavior/About) disables gestures,
but the existing LaunchedEffect only snapped the drawer closed when it was
already isOpen. The drawer's scrim surface lingered during the transition and
intercepted the first back-button tap, sliding the sidebar open instead of
navigating back.
Now snapTo(Closed) unconditionally whenever gestures are disabled — a safe
no-op when already closed — so the sub-page back button works on the first tap.
* [bug] fix settings sub-page back button opening drawer (gesture-surface race)
The first "back" tap on a settings sub-page opened the sidebar instead of
navigating back. Root cause: Settings root is a gesture-enabled drawer screen,
and the sub-pages are the first destinations ever drilled down from it. When
navigating in, ModalNavigationDrawer's gesture interceptor surface lingered
for a frame (gesture -> non-gesture transition), so the top-left tap was
consumed by the drawer instead of the Back IconButton. The previous snapTo
fix only reset drawer *state*, not the gesture layer, so it didn't help.
Fix: NavigationController now holds a closeDrawer hook (set by MainNavigation)
and invokes it synchronously inside navigateTo() when pushing a non-primary
sub-page. The drawer is dismissed in the same tap handler that triggers the
navigation, before any recomposition, so the sub-page is never tappable while
the gesture surface is still mounted. The LaunchedEffect(gesturesEnabled)
snapTo remains as a backstop for non-navigateTo entries.
* fix(#603): delay drawer gesture activation to prevent residual swipe
* fix(#603): fix settings back navigation and drawer auto-open
---------
Co-authored-by: hy4ri <hy4ri@users.noreply.github.com>
Add a Language selector to Settings → Appearance that overrides the app
display language independent of the device locale. Persists a BCP-47 code
in ServerStoreState.appLanguage; MainActivity.attachBaseContext applies it
via a ContextWrapper (works on minSdk 26, no AppCompat dependency needed
since the Activity is a plain ComponentActivity). Selecting a language
recreates the activity so resources re-resolve immediately.
Includes System default / English / 한국어 options, plus a unit test for
the locale-code resolver.
Note: the new string keys (settings_item_language, language_system,
language_english, language_korean) should be backfilled into values-ko
once PR #602 lands.
Co-authored-by: hy4ri <hy4ri@users.noreply.github.com>
Allow the chat input to be sent while the agent is mid-turn or awaiting
approval. The backend's prompt.submit busy-input policy already queues any
prompt that lands during a running turn (tui_gateway/server.py:_handle_busy_submit)
and drains it as the next turn, so the message is never dropped.
- ChatInputPolicy: pure, testable send/placeholder decision logic
- ChatScreen: canSend stays enabled while typing/approving; placeholder shows
'Queued — sends when agent is free' when text is present during a busy turn
- ChatInputPolicyTest: truth-table covering typing, approval-pending,
disconnected, empty, attachment, and slash-command cases
- strings: chat_input_placeholder_queue
Note: this supersedes the Genesi599 fork's approach (dispatchViaRpc('queue ...')),
which double-submits and relies on a 4018-prone command.dispatch path. The
backend-queued prompt.submit path is the correct, single-submission contract.
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#589): show in-session model picker when user sends /model
Typing a bare '/model' in the chat composer now opens the model picker so
the user can hot-swap the CURRENT session's model via the existing
slash-command path, instead of hand-typing the provider/model string.
- ChatViewModel.preloadModelOptions() caches GET /api/model/options at
GatewayReady so the picker opens instantly; falls back to on-demand
refresh if opened before preload finishes.
- sendMessage() intercepts a bare '/model' (no arg) -> showModelPicker=true
instead of dispatching; a typed '/model provider/model' still dispatches
via handleSlashCommand.
- sendSlashModel() routes through command.dispatch with session_id, so the
swap is session-scoped (not the global POST /api/model/set).
- Extracted ModelPickerDialog into ui/model/components and reused by both
ModelScreen and ChatScreen (adds loading + empty states).
- Added a top-bar model chip (current session model label) that opens the
same picker.
- Added unit tests: bare /model opens picker, picker selection hot-swaps via
slash dispatch with session_id, typed /model with arg dispatches directly.
Note: CI shows 2 pre-existing test failures (ChatViewModelTest.testToggleSearch,
HermesWsClientTest.testAutoReconnect) that also fail on a clean main full
suite (MockK static-mock leakage between classes). Unrelated to this change;
compile + ktlint + the 3 new tests pass.
* fix(#589): use backend-valid session-scoped /model command syntax
The model picker now dispatches the canonical backend command form
'/model <model> --provider <slug> --session' instead of the unsupported
'provider/model' slash form. The '--session' flag keeps the switch scoped
to the current chat only (per-session override) and never writes the global
model config, which matches the backend model-switch contract.
- sendSlashModel builds '/model $model --provider $provider --session'.
- Updated unit tests to assert the corrected arg payload
('gpt-4o --provider openai --session') and the fully-typed dispatch path.
* fix(#589): add pinned-models section to in-session model picker
- ModelPickerDialog now renders a Pinned section (mirrors global model screen)
- search filter matches model names too, not just provider name/slug
- ChatUiState gains modelPickerPinned, fed from AuthManager.getPinnedModels()
- top-bar model chip removed; picker opens on /model only
- test mock for getPinnedModels() added
* fix(#589): route /model via prompt.submit instead of command.dispatch (resolves 4018)
/model is a real backend slash command handled in the backend's message
pipeline (_handle_message -> _handle_model_command), so it must arrive as
a normal prompt.submit message, not the command.dispatch RPC (which only
knows quick/plugin/bundle/skill commands and 4018s on /model).
- SlashCommandDispatcher: add SlashResult.ModelSwitch, map /model -> ModelSwitch
- ChatViewModel: handle ModelSwitch -> handleModelSwitch() sends via
WsMethods.PROMPT_SUBMIT with {session_id, text}
- Tests: rewrite the two /model tests to assert prompt.submit (previously
asserted the buggy command.dispatch path); add /model -> ModelSwitch test
Independent agy review: APPROVE, no blocking issues.
* fix(#589): route /model via slash.exec (correct transport; prompt.submit made LLM treat it as text)
Earlier attempt routed /model through prompt.submit, but the TUI gateway's
prompt.submit handler does NOT parse slash commands — it runs the text as a
normal agent turn, so the LLM received '/model ...' as a user message (the
exact reported bug). The correct RPC is slash.exec, which runs the command
through the slash worker and _mirror_slash_side_effects -> _apply_model_switch
to hot-swap the live session model. command.dispatch is also wrong (only knows
quick/plugin/bundle/skill commands -> 4018).
- WsMethods: add SLASH_EXEC = "slash.exec"
- ChatViewModel.handleModelSwitch: send slash.exec with {command, session_id}
(previously prompt.submit with {text}); corrected doc comment
- SlashCommandDispatcher: corrected comments (slash.exec, not prompt.submit)
- ChatViewModelTest: rewrite both /model tests to assert slash.exec with the
/model <model> --provider <slug> --session command string
- SlashCommandDispatcherTest: /model -> ModelSwitch already covered
Verified against hermes-agent tui_gateway/server.py: slash.exec (L13190) ->
worker.run + _mirror_slash_side_effects (L13299) -> _apply_model_switch (L13118);
parse_model_flags accepts --provider/--session (hermes_cli/model_switch.py:355).
Independent agy review of the prior attempt: APPROVE (the wrong-transport
assumption was the author's, not agy's).
* fix(#589): route /model via config.set key=model (slash.exec 4001'd; prompt.submit made LLM treat it as text)
Third transport attempt, now verified against the gateway source:
- prompt.submit: gateway runs text as a normal agent turn -> LLM sees '/model'
as a message (the original reported bug). WRONG.
- slash.exec: gateway returns 4001 'session not found'. The slash worker path
resolves the session differently / requires agent-built state the mobile
session isn't in at switch time. WRONG.
- config.set key=model: gateway (server.py config.set L10253) routes key=='model'
straight to _apply_model_switch using the SAME _sessions.get(session_id)
lookup that the working command.dispatch uses. CORRECT.
sendSlashModel still builds '/model <model> --provider <slug> --session';
parse_model_flags (hermes_cli/model_switch.py:355) extracts model/provider/
session from it. handleModelSwitch now sends config.set {key:model, value:
command, session_id}.
- WsMethods: add CONFIG_SET = 'config.set'
- ChatViewModel.handleModelSwitch: config.set instead of slash.exec
- SlashCommandDispatcher: comments updated (config.set, not slash.exec)
- ChatViewModelTest: both /model tests assert config.set key=model + command
- also fixes the MockK testToggleSearch flake by preloading byte-buddy agent
via JAVA_TOOL_OPTIONS locally (container blocks JVM self-attach)
* fix(#589): strip /model prefix before config.set key=model (5001 'model names can't have spaces')
config.set key=model passes the value DIRECTLY to parse_model_flags — it does
NOT strip a leading '/model' the way slash.exec / prompt.submit do. Sending
'/model <model> ...' made parse_model_flags treat '/model <model>' as the
model name (contains a space) -> 5001.
handleModelSwitch now strips the '/model' prefix and sends the bare spec
'<model> --provider <slug> --session', matching the TUI client's
modelValueForConfigSet (ui-tui session.ts:23) used by useMainApp.ts:123.
- ChatViewModel.handleModelSwitch: spec = command.removePrefix('/model').trim()
- ChatViewModelTest: assert bare spec 'gpt-4o --provider openai --session'
- comment updated
* fix(#589): strip /model prefix case-insensitively before config.set
SlashCommandDispatcher and isModelPickerCommand match /model
case-insensitively, but handleModelSwitch used removePrefix("/model")
(case-sensitive). A typed /MODEL ... forwarded the literal slash prefix
to config.set and the backend's parse_model_flags wouldn't recognize it
— silent hot-swap failure.
Use startsWith("/model", ignoreCase=true) + substring(6) so any casing
is stripped. Added a regression test asserting /MODEL does not forward a
leading slash.
* fix(test): inject searchDispatcher to fix MockKException in testToggleSearch
---------
Co-authored-by: M57 <m57@hermes-control.app>
* feat(#530): add SessionSearch models + searchSessions endpoint + contract tests
* feat(#530): wire server-backed search into Sessions screen (debounce + render)
* style(#530): ktlint import ordering on HermesApiService
* fix(#530): ktlint-clean overlap wrap + map search snippet to title
* fix(#530): map search snippet to title so results show + highlight
* fix(#530): render search hits as honest result cards instead of faking title
Backend SessionSearchResult carries no session name (only snippet + metadata),
so mapping snippet->title made the card show raw excerpt as the title. Instead,
render a dedicated SearchResultCard: 'Match' label + session id, highlighted
snippet as body, and source/model/played-at chips. Added SessionInfo.model and
a friendly relative-time formatter. ktlint clean, compileDebugKotlin + unit
tests pass.
* fix(#530): strip JSON from search snippet + tighten result card padding
* fix(#530): strip FTS highlight markers from search snippet on client
agy's backend change re-wraps matched terms in >>>term<<< markers inside the
clean snippet, but the mobile highlights by raw query term and never consumed
those markers — so literal '>>>'/'<<<' would render in result text. Strip the
markers before passing the snippet to highlightText so the query highlight
still works on clean text.
* fix(#530): integrate tree-flattening hierarchy with search modes in SessionsScreen
* fix(#530): fix test assertion and mock ApiClient in SessionsViewModelTest
* fix(#530): resolve compose PaddingValues compile error and fix slash-preservation test path
* fix(#530): clean search snippet from JSON payload and add unit test
---------
Co-authored-by: M57 <m57@hermes-control.app>
The backend terminal tool (tools/terminal_tool.py) returns a single merged
'output' string plus 'exit_code' and 'error' — never 'stdout'/'stderr'.
ChatBubble was reading the non-existent 'stdout' key, so command output was
silently dropped and only the Duration footer showed.
- Parse 'output' as canonical; keep legacy 'stdout'/'stderr' fallback for
backward compat.
- Route terminal output into mainOutput so it renders like other tool bubbles.
- Surface non-zero exit codes as an error line when no 'error' string is set.
- Remove dead 'stderr' render branch.
- Update ToolBubbleParsingTest to assert the real payload shape (incl. the
exact issue-585 'echo hellp' sample) and add regression cases.
Co-authored-by: hy4ri <hy4ri@users.noreply.github.com>
* refactor(theme): split per-preset color schemes into theme/presets/
Move each ThemePreset's ColorScheme + status colors into its own file under
theme/presets/ (Default, Monochrome, Gruvbox, Catppuccin, Amoled, NeonNoir).
Theme.kt becomes a thin dispatcher via resolveColorScheme/resolveStatusColors,
making individual themes far easier to create and edit.
- AMOLED keeps a dark-only scheme (falls back to default light)
- MONOCHROME/AMOLED reuse default status colors (no bespoke semantic set)
- aligned to current Voltage Purple palette + on* status-color params
Verified: ktlint clean, compileDebugKotlin passes.
* refactor(theme): make all preset files uniform 4-token shape
Every preset file now exports XxxDarkColorScheme / XxxLightColorScheme /
XxxDarkStatusColors / XxxLightStatusColors so they share one shape and import
order. MONOCHROME + AMOLED alias the default status colors; AMOLED also gets an
explicit AmoledLightColorScheme = DefaultLightColorScheme alias.
Theme.kt dispatcher is now a pure lookup (no special-casing) for both color
scheme and status colors. No runtime behavior change — pure structural cleanup.
Verified: ktlint clean, compileDebugKotlin passes.
* docs(theme): add THEMES.md documenting the per-preset convention
Co-located in theme/ — covers layout, the uniform 4-token shape, the pure
lookup dispatcher, why DEFAULT uses design tokens (intentional, not a bug),
and a step-by-step 'add a new theme' recipe.
Verified: ktlint clean.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#584): follow stream without freezing (persistent snapshotFlow)
The auto-scroll LaunchedEffect was keyed on (messages.size,
streamingMessage?.content?.length, isThinking) — those change on EVERY
streamed token, so the effect restarted per token and cancelled the
in-flight animateScrollToItem, leaving the view stuck/frozen mid-stream.
Replace it with a single persistent LaunchedEffect(Unit) + snapshotFlow +
collectLatest. The streaming follow now uses an instant scrollToBottom
(animated=false) which is idempotent per token and has no animation to
cancel; a discrete new message arriving while at the bottom still animates.
Because the effect runs once, mirror the params through rememberUpdatedState
and read them live inside snapshotFlow/collectLatest so the flow actually
fires per token and session-switch detection uses the current session id
(rather than stale first-composition closure captures).
* fix(#584): leave scroll free during streaming (no auto-follow)
m57 follow-up: the previous fix stopped the freeze but still force-pinned
to the bottom while streaming. Now, while a message is streaming, the
scroll is left completely free — no auto-follow of the streaming message —
so the user can scroll up/down to read while it generates.
Auto-scroll only on explicit actions (send / FAB / session switch) and when
a discrete non-streaming message lands while already pinned to the bottom.
streamingMessage is dropped from the snapshotFlow keys so the flow no
longer churns per token.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#583): align last message bottom to viewport on scroll-to-bottom
animateScrollToItem(lastIndex) only top-aligns the last item, so when the
last message is taller than the viewport its bottom is clipped below the fold.
Replace the 4 scroll-to-bottom call sites (send auto-scroll, session-switch,
auto-scroll-on-new-message, and the FAB) with a scrollToBottom() helper that
top-aligns then scrolls the exact remaining delta so the last item's bottom
edge is visible.
Using the remaining delta (not scrollOffset=Int.MAX_VALUE) avoids integer
overflow in the internal scroll-position clamp, which would wrap the offset to
0 and scroll back to the top for tall items.
* fix(#583): keep smooth animation on scroll-to-bottom when animated
agy review (PR #590) flagged that the animated path did an instant
scrollToItem then only animated the small remaining delta, causing a hard
jump on FAB / send clicks instead of a smooth scroll. Use animateScrollToItem
when animated, scrollToItem when instant.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#587): forward session_id on image.attach_bytes and await the RPC
The mobile image-send path called image.attach_bytes without the required
session_id field, so the gateway resolver (_sess) returned
JsonRpcError(code=4001, message=session not found).
- Add "session_id" to agentSessionId to the image.attach_bytes params
(desktop reference client already does this; mobile was the outlier).
- Switch the image branch from fire-and-forget wsClient.send to the
awaited sendRpcAndAwait() layer (same awaited-RPC path used for
slash commands, #526). This also removes the race where prompt.submit
could fire before the image was staged, and surfaces RPC errors
instead of silently swallowing them.
Root cause verified against tui_gateway/server.py _sess() and the
desktop use-prompt-actions reference. ChatViewModel.kt passes ktlintCheck.
Closes#587
* fix(#587): also forward session_id on file.attach (same 4001 bug)
The image branch was fixed in the prior commit, but the file branch
(sendMessage else -> file.attach) omitted session_id too, so non-image
file uploads also hit JsonRpcError(code=4001, "session not found").
The backend file.attach handler (tui_gateway/server.py _sess) requires
session_id exactly like image.attach_bytes.
Add "session_id" to agentSessionId to the file.attach params, and
refresh the sendMessage flow comment (no longer fire-and-forget).
Verified by source: both image.attach_bytes and file.attach call the
same _sess resolver. Camera capture is unaffected (it produces an
image/jpeg attachment, which already routes through the fixed
image.attach_bytes branch).
ChatViewModel.kt passes ktlintCheck.
* test(#587): regression for image.attach_bytes session_id
Adds testSendMessage_imageAttachment_sendsSessionId, which proves the
image-attach RPC is issued with session_id (the missing field that
caused JsonRpcError 4001 'session not found').
Root blocker found: readContentUriBase64 constructs
android.util.Base64OutputStream, which throws MockK's 'Stub!' on the
JVM test, making the read return null and the image branch never run.
Mocking the Base64OutputStream constructor lets the read succeed so the
request(any(), capture(paramsSlot), any()) stub captures the params and
the session_id assertion holds.
Full ChatViewModelTest suite passes; ktlint-clean.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#575): drop description text from slash-command suggestions
* fix(#575): show /stop and /interrupt in slash-command menu
* fix(#574): hide desktop/CLI-only slash commands from mobile suggestion menu
Recreate hiddenSlashDisplay blocklist removed in 1f322e3 with the 33
cli_only/TUI-only commands that don't function on mobile (issue #574).
/stop and /interrupt intentionally remain visible (they work on mobile, #575).
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Co-authored-by: M57 <m57@hermes-control.app>
* fix(#575): drop description text from slash-command suggestions
* fix(#575): show /stop and /interrupt in slash-command menu
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* fix(#572): clarify Dismiss now informs agent instead of silently dropping
* fix(#572): clarify Dismiss rejects the question (no answer) instead of telling agent to proceed
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix: refresh loopback dashboard token after restart
The gateway regenerates its ephemeral session token on every restart, so the
token the app persisted becomes stale. After a dashboard restart REST calls
401 and the WebSocket auto-reconnect loop hangs in AUTH_EXPIRED until the user
manually re-authenticates.
- Add DashboardSessionTokenRefresher: re-reads the injected
window.__HERMES_SESSION_TOKEN__ from the dashboard SPA and writes it back to
AuthManager.
- TokenRefreshAuthenticator: on a 401 retry, scrape a fresh token from the
dashboard and retry instead of giving up.
- HermesWsClient: refresh the token before each non-gated WebSocket handshake so
reconnects don't get stuck on a stale token; guard serverStore read against
IllegalStateException.
Covers the loopback (--insecure) deployment path; gated/ticket mode is untouched.
* fix: harden dashboard token refresher (sync, body cap, blank-token test)
Second-opinion review flagged robustness gaps (no correctness/security
regressions in the happy path):
- DashboardSessionTokenRefresher.refresh(): @Synchronized to dedupe
concurrent 401 / reconnect storms instead of each firing its own GET.
- Cap SPA HTML read at 64 KiB before regex scan (defensive against OOM on
low-end devices); we only need the injected token near the document top.
- HermesWsClient: log (not silently swallow) the transient uninitialized
serverStore case so a real misconfig is still visible.
- Add test: blank-injected token returns null.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* feat(#538): wire tool.progress/generating + subagent.* events into chat
* fix(#538): remove subagent indicator on subagent.complete so it doesn't linger
* style(#538): drop dead Elvis on non-nullable event.text in onMessageComplete
---------
Co-authored-by: M57 <m57@hermes-control.app>
* feat(#533): add session.branch (fork a conversation) to mobile
The gateway already supports session.branch (tui_gateway/server.py), but
the mobile had no client surface. This adds:
- WsMethods.SESSION_BRANCH = "session.branch"
- A Branch (CallSplit) action on each session card that forks it via the
session.branch WS RPC with the source session_id (and optional name).
- branchedSessionId state -> SessionsScreen navigates to the freshly
forked session on success; RpcError -> toast; spinner while branching.
- SessionsViewModelTest (4 tests): forwards session_id, forwards name,
sets branchedSessionId on success, toast+no-branch on RpcError, and
consumeBranchedSession clears the pending id.
Verified against the live gateway contract: session.branch expects
{session_id, name?} and returns {session_id, title, parent}. Local
testDebugUnitTest: 4/4 pass, ktlint clean.
* refactor(#533): make /fork work via chat instead of a session-menu button
Per review feedback: the session.branch surface belongs in the chat
composer, not the session list menu.
- SlashCommandDispatcher: /fork and /branch -> new SlashResult.SessionBranch.
- ChatViewModel: handles SessionBranch by sending session.branch with the
active runtime session id (optional arg = branch title), and on the RPC
result switches the current session to the new forked one (mirrors the
SESSION_CREATE handler). This fixes /fork, which previously fell through to
command.dispatch and 4018'd (backend only knows session.branch).
- Removed the Branch (CallSplit) button + navigation LaunchedEffect from
SessionsScreen/SessionsViewModel (reverted those two files to main) and
dropped the now-unused sessions_action_branch string.
- Tests: deleted SessionsViewModelTest (branch-only); added /fork -> SessionBranch
to SlashCommandDispatcherTest and two ChatViewModel tests asserting /fork
sends session.branch with the session id + optional name.
Local testDebugUnitTest: SlashCommandDispatcherTest 8/8, ChatViewModelTest 50/50,
0 failures. ktlint clean.
---------
Co-authored-by: ronia <ronia@m57>
* fix(#548): stop System tab button text truncation + fix two wrong labels
* fix(system): remove inline icons from narrow system action buttons to prevent text cutoff
* fix(system): restore action icons and implement dynamic font scaling for button text
* fix: stack icon above text in system buttons to prevent label clipping
Replace horizontal Icon+Text layout with vertical Column stacking
(icon on top, text centered below) — the same pattern Material 3
NavigationBar items use. Gives text the full button width so labels
like 'Security Audit' and 'Config Migrate' never get cut off.
* fix: add missing ImageVector import for system buttons fix
---------
Co-authored-by: M57 <m57@hermes-control.app>
Mobile was silently dropping two event types the gateway emits. This adds
parsing + UI surfacing for both, bringing mobile to parity with desktop
(which shows gateway errors as red toasts).
- `error` (backend crash) -> WsEvent.GatewayError, surfaced in the existing
error banner (mirrors how RpcError is handled).
- `background.complete` (scheduled job finished) -> WsEvent.BackgroundComplete,
surfaced as a non-blocking snackbar in ChatScreen.
Changes:
- WsEvent.kt: add GatewayError(message?) + BackgroundComplete(data?) sealed subtypes.
- EventParser.kt: handle "error" + "background.complete"; unknowns still fall
through to Unknown (no behavior change for existing event types).
- ChatWsEventReducer.kt: onGatewayError (sets errorMessage, clears loading) +
onBackgroundComplete (sets backgroundCompleteMessage); mirrors onRpcError.
- ChatViewModel.kt: backgroundCompleteMessage state field, clearBackgroundComplete(),
handleWsEvent branches (ViewModel owns the snackbar trigger).
- ChatScreen.kt: pass backgroundCompleteMessage to ChatLifecycleEffects + a
LaunchedEffect that shows the snackbar and calls clearBackgroundComplete().
- EventParserTest.kt: cover both new events (incl. missing-message case).
Pure additive; sealed class + exhaustive when() preserved. Verified:
compileDebugKotlin + ktlintCheck green; EventParserTest passes.
Closes#527
Co-authored-by: M57 <m57@hermes-control.app>
* fix(ws): align RPC timeout to 120s + rejectAllPending on disconnect (fixes#526)
Mobile's pending-request handling diverged from the desktop contract
(apps/shared JsonRpcGatewayClient) in two ways that caused silent hangs:
1. Timeout was 30s while the desktop uses 120s (DEFAULT_REQUEST_TIMEOUT_MS),
so legitimately long agent turns were pruned early. Align to 120s and use
it for both the cleanup sweep and sendRpcAndAwait's per-call withTimeout.
2. The cleanup sweep only Log.w'd stale requests and dropped them — it never
failed the CompletableDeferred, so sendRpcAndAwait callers awaiting a
server that never answered HUNG FOREVER. And on disconnect/reconnect there
was NO rejectAllPending, so in-flight awaited RPCs also hung until their
own timeout.
Changes (all in ChatViewModel.kt):
- cleanup sweep now completeExceptionally()s the deferred AND surfaces a
'Request timed out' error for fire-and-forget RPCs (was silently dropped)
- new rejectAllPending() fails + clears every in-flight pending deferred;
wired into the disconnect-status collector and reconnect()
- PENDING_REQUEST_TIMEOUT_MS = 120_000 (matches desktop); sendRpcAndAwait
uses it instead of a hardcoded 30s
Tests: 2 new ChatViewModelTest cases cover rejectAllPending on disconnect
and on reconnect (no hang, VM stays usable). (Timeout-surfacing path shares
the deferred-fail logic with handleRpcError; the 120s sweep can't run under
the unit-test dispatcher's periodic-timer limitation, noted in PR.)
Mirrors the desktop-app audit in /opt/hermes-agent.
* fix(ws): aggregate pending-request cleanup into a single UI update (PR #541)
Sourcery review (PR #541): startPendingRequestCleanup emitted one
_uiState.update{} per stale request inside the loop, causing N separate
emissions / redundant recompositions when many RPCs time out at once.
Collect fire-and-forget methods and emit a single combined errorMessage
update after the loop.
rejectAllPending is intentionally left silent on disconnect/reconnect:
those are transient socket events (the RPC often completes on the new
socket), so surfacing a scary error banner is UX noise. Only the
timeout path surfaces errors — that is a genuine failure. This matches
the desktop JsonRpcGatewayClient.rejectAllPending semantics.
* refactor(ws): move awaited-RPC pending/timeout layer into HermesWsClient (closes#526)
Issue #526: port the JsonRpcGatewayClient.request() pattern into
HermesWsClient. HermesWsClient now owns the in-flight map, per-call
120s timeout, resolve-on-RpcResult/RpcError, and rejectAllPending on
disconnect/reconnect. ChatViewModel.sendRpcAndAwait simply delegates
to wsClient.request(...).await(); its local pendingRequests map,
PendingRequest, startPendingRequestCleanup loop and rejectAllPending
are removed. Fire-and-forget send() is unchanged.
---------
Co-authored-by: M57 <m57@hermes-control.app>
* fix(network): scope management REST calls with ?profile= (fixes#528)
Mobile fired all management REST calls with NO profile scope, so every
profile read/wrote the backend's implicit default profile — multi-profile
setups silently corrupted config/skills/toolsets/mcp/model/env.
Add ProfileScopeInterceptor (OkHttp) that appends ?profile=<activeProfile>
to the same endpoint families the desktop/web dashboard scopes
(web/src/lib/api.ts PROFILE_SCOPED_PREFIXES), respecting:
- no active profile -> pass through (legacy default behavior)
- explicit ?profile= on the URL wins over the global scope
- only scoped families rewritten; ops/pairing/cron/profiles stay global
Wired into ApiClient.buildService(). 5 MockWebServer tests cover scoped
append, explicit-param precedence, no-profile passthrough, non-scoped skip,
and skills scoping.
Mirrors desktop/web contract derived from audit of /opt/hermes-agent.
* fix(ci): stub getSelectedProfileId in ApiClientTest + harden prefix match (fixes PR #540)
- ApiClientTest: stub AuthManager.getSelectedProfileId() to null. The new
ProfileScopeInterceptor (registered in ApiClient.buildService) calls it, and
the unstubbed real method threw IllegalStateException at AuthManager.kt:56,
breaking all 4 auth-interceptor tests.
- ProfileScopeInterceptor: replace startsWith(prefix) with whole-segment match
(path == prefix || path.startsWith(prefix/)) so lookalike paths like
/api/statusXYZ are NOT wrongly scoped (Sourcery review, PR #540).
- ProfileScopeInterceptorTest: add lookalikePath_notScoped + scopedSubPath_isScoped
to lock the segment-boundary fix.
Previously mobile dropped the gateway's sudo.request and secret.request
events, so any agent turn needing a sudo password or a secret value hung
forever with no UI. Now:
- EventParser maps sudo.request / secret.request -> WsEvent.SudoRequest / SecretRequest
- ChatViewModel surfaces a secure dialog (SudoPromptUi / SecretPromptUi) and
replies via sudo.respond / secret.respond with session_id + request_id + payload
- ChatWsEventReducer explicitly branches the new events (no-op, VM-owned)
- ChatScreen renders password-masked SudoPromptDialog / SecretPromptDialog
- 6 unit tests cover prompt-state, RPC params, and prompt-clearing
Mirrors desktop/web contract (apps/shared GatewayEventName + tui_gateway).
Derives from audit of /opt/hermes-agent desktop app.
Explicit IO context for DB/WS/notify path. replyScope already uses
Dispatchers.IO; this makes the offload local and explicit. Keep return@withContext
only because nothing follows the withContext block today.
The parseToolOutput function in ChatBubble.kt was handling the specific
parsing and formatting logic for the "todo" tool inline, making the
function long and harder to read.
This commit extracts the todo-specific formatting logic into a new
private function formatTodoToolOutput, improving the readability and
maintainability of ChatBubble.kt without changing any behavior.
Refactored the deeply nested `parseDaysOfWeek` function using guard clauses to flatten the logic structure. By returning early on invalid conditions (`dowPart == "*" || domPart != "*" || monthPart != "*"` and `hoursList.size != 1 || minutesList.size != 1`), the remaining function body's indentation is reduced by two levels.
This improves overall code readability and maintainability without altering existing functionality. The change was verified against existing unit tests, which all pass.
The hub card UI is rendered by the local private HubSkillCard in SkillsScreen.kt (not the unused components/HubSkillsTab.kt copy). Moved SourceBadge out of the top-right name row to the left of the Install button in the card footer (left-aligned source, right-aligned action), tying the skill origin (hub/clawhub/etc) to the install action.
* feat(#500): tap-to-reveal detail dialog for truncated cards
Add shared DetailDialog composable and wire tap-to-open full metadata
dialogs across Skills (installed + hub), MCP servers, Plugins, Toolsets,
and Webhooks screens. Action buttons stay on the card (sticky); tapping
the card body reveals untruncated description + key metadata.
Refs: #500
* fix(#500): resolve compile errors from missing runtime imports + dialog scope
- add missing androidx.compose.runtime.mutableStateOf/remember imports
in InstalledSkillsTab + HubSkillsTab (caused 'Unresolved reference'
cascade on compileDebugKotlin)
- move PluginsScreen detail dialog inside the composable function scope
(was nested in OrphanPluginCard -> showDetail unresolved)
- DetailDialog rows type: Pair<String?,String?> -> Pair<String,String?>
(key always non-null; fixes type-inference failure)
- drop unnecessary safe call on non-null label
Verified locally with the Android SDK:
./gradlew :app:compileDebugKotlin -> BUILD SUCCESSFUL
./gradlew :app:testDebugUnitTest -> BUILD SUCCESSFUL
Refs: #500
* refactor(#500): extract toDetailRows() mappers + semantic status tones; wire SkillsScreen hub browse
- add DetailRows.kt with @Composable extension mappers for Skill/HubSkill/
McpServer/PluginInfo/Toolset/WebhookSubscription -> List<DetailRow>
- DetailDialog now renders status rows via LocalHermesStatusColors (review #2):
SUCCESS/NEUTRAL/WARNING/ERROR/INFO tones instead of plain onSurfaceVariant
- wire SkillsScreen.HubBrowseView/HubSkillCard tap -> DetailDialog (same UX as
other screens) — user request
- remove JVM-signature-clashing Pair overload; drop unused Color import
- add detail_dialog_trust_level string (HubSkill has no version field)
- verified: compileDebugKotlin + testDebugUnitTest PASS locally w/ ANDROID_HOME
* fix(#500): full hub description in dialog + no name truncation
- HubBrowseView (both SkillsScreen + HubSkillsTab): on card tap, fetch
previewHubSkill(identifier) to get the FULL SkillContentResponse.content
(search snippet was truncated); dialog shows full content once loaded,
with a spinner while previewing. Falls back to snippet if no identifier.
- SkillsViewModel: add hubPreviewContent/isHubPreviewing/hubPreviewError
state + previewHubSkill()/clearHubPreview() (uses existing preview endpoint)
- DetailDialog: title now wraps fully (removed maxLines=2 ellipsis cap) so
long skill names are never cut off
- verified: compileDebugKotlin PASS + ktlint clean (local ANDROID_HOME)
Backend commit 37a4cf900 (hermes-agent) flipped the default of
GET /api/model/options's include_unconfigured parameter from true
to false, so the mobile Models picker was silently regressing to
showing only configured/authenticated providers — the skeleton rows
for unconfigured providers (with setup-hint warnings) stopped arriving.
Add include_unconfigured as a @Query param defaulting to true so the
full provider universe is restored on the Models screen. Both current
call sites (ModelViewModel + ProfilesViewModel) invoke the API without
passing the new param, so they automatically pick up the true default.
