* fix(chat): preserve reasoning across tool calls + REST reload (#771)
A reasoning-model turn streams its thinking BEFORE the tool starts. The
reducer wiped the streaming state at tool.start (returned a fresh
StreamingState()), so the reasoning bubble vanished mid-turn and the
finalized answer had no reasoning card.
Real gateway capture for `run echo hi` (reasoning model):
message.start
reasoning.delta xN (thinking before the tool)
tool.generating/start/complete (terminal)
message.delta xN (answer after the tool)
reasoning.available (FULL trace in payload text)
message.complete (text + reasoning field)
Fixes:
- onToolStart keeps the streaming message + reasoningText alive across
the tool call (only buffers are cleared via clearStreamingBuffers);
an interim-text orphan still seals, but reasoning survives for the
post-tool answer
- EventParser now reads reasoning.available text and message.complete
reasoning (the authoritative full trace the gateway sends)
- onReasoningAvailable fills the streaming message with the full trace
- onMessageComplete prefers the payload reasoning over delta leftovers
- mapServerMessages folds reasoning-only empty assistant rows into the
next real answer bubble (no more standalone reasoning ghost bubble)
and carries tool names over from the live WS list by position (no
more generic 'tool' title after a REST reload)
Regression tests replay the captured sequence end-to-end.
* fix(chat): dedupe WS + REST tool bubbles by result content (#771)
The REST transcript tool rows carry no tool name, and WS messages (UUID
ids) + REST rows (rest- ids) are BOTH persisted to Room — the cache
paint and reloads could render the same tool call twice, with the
second bubble falling back to the generic 'tool' title.
Real captured pair for 'echo meow':
terminal {tool_id, name, args, result:{...}} (WS)
Tool {output, exit_code, error} (REST, no name)
The previous positional carry-over misaligned whenever the REST page
and the live tool list didn't line up 1:1 (multiple tool calls in
history → newest call got null name).
Fixes:
- canonicalToolResultKey(): content-based matching between the WS
tool.complete payload and the REST result row — int/float agnostic
(0 vs 0.0), prefers the payload's 'result' field
- mapServerMessages(): when a REST tool row's result matches a live WS
tool message, REUSE the live message wholesale (same id + toolName +
rich payload) so Room upserts instead of accumulating a rest- copy
- dedupeCachedMessages(): cache paint drops rest- copies whenever a WS
copy of the same logical message exists (tools matched by result,
other roles by content)
11 new unit tests with the real captured payloads.
* fix(chat): merge transcript reload with live WS bubbles (#771)
The gateway persists a tool row only when the tool COMPLETES server-side.
A REST transcript reload landing mid-turn (app background/foreground,
reconnect re-resume, pull-refresh) therefore returned a page WITHOUT the
in-flight tool row — replacing the message list outright made the tool
bubble vanish and left tool.complete with no RUNNING message to update.
- mergeTranscriptWithLive(): reload now merges instead of replacing —
any current message the REST page doesn't cover (by id or logical
content) survives, chronological order preserved
- loadSessionMessages applies the merge before publishing state
4 new unit tests: in-flight RUNNING tool survives a mid-turn reload,
reused-id and canonical-match cases don't duplicate, order preserved.
* test(chat): end-to-end tool-bubble survival through mid-turn reload
Chains the real user scenario at the state level: WS turn (message.start →
reasoning → tool.start → tool.complete) + a transcript reload landing
while the tool is still running server-side (REST page lacks the tool
row) + stream finish. Asserts the tool bubble survives to the final
list with its name and status intact.
* debug(chat): add Logcat tracing for tool-bubble vanish (#771)
Temporary diagnostic logging to catch the exact moment and code path
where the tool bubble disappears. Logs to tag 'ChatReducer771' and
'ChatVM771' at every state mutation that touches messages:
- onMessageComplete: message count + roles before the assistant append
- loadSessionMessages: rest vs current vs merged counts and roles
- syncCurrentSession: trigger + current tool count
- sync merge: before/after merge counts and roles
To reproduce: install this debug APK, run 'echo meow', watch the tool
bubble vanish, then capture logcat:
adb logcat -d -s ChatReducer771:D ChatVM771:D
Will remove the logging once the root cause is pinned.
* fix(chat): drop debug logging + harden sync merge (#771)
The debug commit added android.util.Log calls in the reducer's
onMessageComplete path, which is a Stub! in JVM unit tests — every
test touching that path died with RuntimeException, and the long
log lines broke ktlint (max-line-length). Remove them.
Sync merge fix: the 5s sync poll could drop the newest WS tool
bubble when the REST page lagged one row behind. Route the merge
through sameLogicalMessage (canonical result-key match) and keep
any WS message that has no REST counterpart.
Adds 2 regression tests for the sync-merge scenario.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Opening a session from history fired session.context_breakdown/session.usage
with the storage session id before session.resume had registered it on the
gateway. Those RPCs resolve against the LIVE runtime registry (_sess_nowait)
and returned 4001 "session not found", which the reducer surfaced as a
spurious snackbar even though the session loaded fine.
- fetchContextUsage() now skips the two live-registry RPCs while
runtimeSessionId is null (pre-resume window); the REST-based context
fallbacks stay live.
- runtimeSessionId is cleared on socket drop so a stale runtime id can't be
reused after a gateway restart/reconnect (same 4001 family).
- regression test: no context/usage RPC fires with the storage id during the
pre-resume window; after the resume result the RPCs fire with the runtime id.
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Bring the desktop's session reconnection UX to mobile:
- Warm-cache fast-path: switchSession paints the cached Room transcript
immediately instead of blanking the screen, then refreshes from the
server in parallel (stale cache can't clobber a fresher server page)
- Bounded auto-retry: a failed session.resume (WS reject or REST
transcript failure) retries with exponential backoff 1s->2s->4s->8s,
capped at 4 attempts; WS + REST failures in the same cycle count as one
- Explicit error recovery: after exhaustion the chat shows an error card
with a manual Retry instead of latching the spinner forever; manual
retry resets the backoff cycle
- Reconnect-aware resume: gateway (re)connect clears stale error state
and re-resumes the current session on the new socket
- Unit tests: warm-cache paint, bounded retry exhaustion, WS+REST
dedupe, manual retry reset, reconnect restart
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat(chat): rewire tool bubbles onto the desktop engine (stage 3)
parseToolOutput is now a thin adapter: it normalizes the payload shape
(args/result sub-objects, legacy top-level fallback, name resolution)
and delegates all per-tool formatting to ToolViewBuilder.
- ToolResultParser.kt: 1325 lines of bespoke formatters → 45-line adapter
- ParsedToolData deleted; ToolBubble renders ToolView directly:
collapsed = emoji + engine title (+countLabel) + subtitle line;
expanded = error → $ command / stdout / stderr / exit code →
DiffViewCard → search hits → detail → duration footer
- Gson-compat shims, formatTodoToolOutput, toIntOrNull and all
per-tool formatters died with the old parser (engine is idiomatic kotlinx)
- Kept mobile-only UI: progress preview, SecurityRiskChip, emoji
prefixes, raw-JSON toggle, DiffViewCard, copy
- ToolViewBuilder fixes found during the rewire: numberValue now accepts
numeric strings (TS Number() semantics — gateway ships exit_code as
string/float), memory title keeps its target suffix
ToolBubbleParsingTest rewritten as adapter-contract tests (14 cases);
per-tool behavior lives in ToolViewBuilderTest. Verified: ktlint +
full testDebugUnitTest.
* fix(chat): dedupe tool name in collapsed bubble summary
The header row already shows the tool name, but the collapsed summary
showed emoji + engine title — and for ~11 tools the engine title is just
the generic humanized name (fact_store -> "Fact Store", todo -> "Todo"),
so the bubble displayed the tool identity twice.
composeToolSummaryLines now drops the title when it equals the generic
humanized name and merges the subtitle into the emoji line instead
("🧠 Fact added (ID: 5)"), restoring the pre-engine one-line style.
Descriptive titles ("Searched \"cats\"", "Ran ls", "Read a.kt L10-14")
stay on line one with the subtitle on line two. A lone emoji with no
subtitle renders nothing.
ToolViewBuilder.genericTitleFor exposes the generic title for the
comparison. 9 new unit tests.
---------
Co-authored-by: hy4ri <hy4ri@users.noreply.github.com>
* feat(chat): port desktop heuristic JSON summarizer engine
New com.m57.hermescontrol.ui.chat.tool.ToolResultSummary — a faithful
Kotlin port of the desktop app's lib/tool-result-summary.ts (verified
against its vitest suite, ported 1:1 to JUnit):
- formatToolResultSummary: wrapper unwrap (data/result/output/response/
payload, up to 4 levels), priority-key ordering, array/record block
formatting with depth caps, noise filtering (success:true etc.)
- extractToolErrorMessage: cycle-safe nested hunt for real error text
through wrappers, with placeholder filtering (error:'none' ignored)
and stderr deliberately excluded as an error signal
Also handles stringified-JSON payloads (re-parse on demand) like the
desktop's tryJson/norm layer.
Ported contract tests: ToolResultSummaryTest (10 cases).
* feat(chat): port desktop per-tool display engine (ToolViewBuilder)
Second half of the desktop engine port, in the new ui.chat.tool package:
- ToolView: the render model the builder produces — status/title/subtitle/
detail/count/duration/stdout/stderr/exitCode/diff/searchHits/imageUrl —
no raw JSON in UI code
- ToolViewBuilder: port of the desktop buildToolView (fallback-model):
status heuristics (explicit success wins; non-zero exit alone is NOT an
error), dynamic titles (Read/Searched/Ran with summarized commands),
per-tool subtitle/detail (browser snapshot stats, terminal output
preview, memory message, cron job counts), count-label cascade,
web_search hit parsing, inline diff + stats, stdout/stderr split,
duration formatting, redundant-detail suppression
- Mobile-specific tool cases carried into the builder: todo, fact_store,
session_search, skills_list/view, skill_manage, process, x_search,
vision_analyze, tool_search, image_generate, project_*, read_terminal,
computer_use
- CommandSummarizer: port of the desktop summarize-command.ts (quote-aware
compound split, pipe-tail strip, redirect/env cleanup)
- Terminal stdout reads the gateway's canonical merged `output` key with
stdout/stderr fallback (mobile terminal_tool contract differs from
desktop)
Tests: ToolViewBuilderTest (18), CommandSummarizerTest (8), all ported
or derived from the desktop suite. Full suite green: 655 tests.
---------
Co-authored-by: hy4ri <hy4ri@users.noreply.github.com>
A new assistant message was seeded with the previous message's
reasoningText (and isReasoning flag) via onMessageStart, and
resetStreaming() never cleared the shared streaming state. If the
next reply produced no reasoning, the stale trace stayed glued to
the new bubble and got persisted to Room.
- onMessageStart: new message + StreamingState start with empty
reasoning instead of inheriting the previous turn's
- resetStreaming: clears isReasoning + reasoningText on the shared
state (finalized messages keep their own copy for the card)
- regression tests: MessageStart seeding, no-reasoning persistence,
reset clearing, finalized-message preservation
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat(chat): add inline video player component and modal viewer
Closes#722
* feat(chat): modernize video player with compact lock pill, save, and share
* refactor(chat): simplify inline video thumbnail to launch fullscreen player on tap
* fix(chat): pass session cookie and auth token headers to VideoView for protected media playback
* fix(chat): format unquoted cookie header for VideoView auth
* fix(chat): resolve video URLs via resolveImageUrl for gateway-hosted media
Both commands are real backend command.dispatch subcommands (name=="init" / name=="focus", hermes-agent 2b0fb72ac). Mobile already routes them generically via RpcDispatch (name/arg/session_id) and handles responses (type:send -> submitPrompt for /init, type:exec -> assistant message for /focus) + surfaces them from commands.catalog.
Adds 6 contract tests mirroring the existing /help + /queue + /status cases:
- /init and /focus forward name+arg+session_id to COMMAND_DISPATCH
- /init with extra arg and /focus on/off forward the arg
- type:send (init) submits the returned prompt via sendMessage
- type:exec (focus) surfaces output as an assistant message
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat(chat): full-screen image viewer (issue #723)
Tap any chat image (markdown , inline attachment thumbnail,
or bubble image) to open a full-screen viewer:
- pinch-zoom + pan via native detectTransformGestures
- downswipe-to-dismiss at min zoom
- Save -> device Downloads (MediaStore), Share -> Android share sheet
(both fully device-local; never sent back to the Hermes server)
Plumbing: onImageClick(ImageViewerModel) flows MarkdownText ->
ChatBubble/InlineAttachment -> ChatMessageList -> ChatScreen (state +
ImageViewerDialog). ImageBytesResolver reuses the existing authed
OkHttpProvider path for gateway http(s) sources; data:/content://
handled locally. No new dependencies.
Closes#723
* fix(chat): resolve strings at composable scope to satisfy LocalContextGetResourceValueCall lint
Lint forbids reading resource *values* from LocalContext.current inside
coroutine lambdas. Hoist all viewer strings via stringResource() at
composable scope and reuse them in onSave/onShare.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* fix(#724): inline host-path MEDIA: images on mobile (client-side)
The previous approach re-fetched the REST session-messages endpoint, but that
endpoint returns the raw MEDIA:<path> text (api_server._message_response does
not inline, and the transcript stores raw content). The desktop app instead
reads the file locally via readFileDataUrl because it is co-located with the
host. Mirror that on mobile: on assistant message.complete/done, inline any
MEDIA:<path> image into a base64 data:image/... markdown tag the renderer can
show. Mobile-only, backend untouched. Unreachable tags are stripped (graceful
degrade). Pure inlining logic extracted to MediaInline for unit testing.
* fix: correct RegexOption name (IGNORE_CASE) in MediaInline
* test: fix byte literal types in MediaInlineTest
* fix(#724): rewrite host-path MEDIA: to gateway /api/files/download URL
Pivot from the earlier base64-inline approach to the desktop-proven,
remote-phone-working fix: rewrite MEDIA:<path> directives (delivered raw over
the WS stream) into authenticated
 markdown tags,
mirroring desktop media.ts mediaExternalUrl. The mobile renderer (MarkdownText)
shows http(s) ![]() via Coil, so agent-sent images now render on mobile too —
including on a remote phone (real HTTP, unlike a host-local file read).
Mobile-only, backend untouched. Pure rewrite logic in MediaUrlRewriter
(replaces MediaInline) with unit tests.
* feat(#724): full gateway-file support — attach MEDIA: files (not just images)
Pivot the MEDIA: fix from image-only URL rewriting to real attachments:
- Attachment: add gatewayUrl + AttachmentSource (LOCAL/GATEWAY) so
gateway-hosted files are first-class and render via existing InlineAttachment.
- HostMediaExtractor (replaces MediaUrlRewriter): parse ANY absolute MEDIA:
directive (any extension) and strip them from message text.
- ChatViewModel.attachHostMedia: build the authed /api/files/download URL via
GatewayFileClient, classify via MediaTypes, attach as Attachment (images
inline, every other file tappable). Idempotent.
- InlineAttachment: tappable. Gateway files fetch via GatewayFileClient and
share via a chooser Intent (FileProvider); images load the URL in Coil.
- Hence agent-delivered images, audio, video, CSV, PDF and arbitrary files all
work on mobile — including a remote phone. Backend untouched.
Builds on #726 (GatewayFileClient). Rebased onto main after #726 merged.
* fix(#724): resolve imports, bubble callback scope, and quoted media path regex
* feat(chat): open attachment in-place on tap (#724)
Tap opens LOCAL (content://) and GATEWAY (fetch->FileProvider) files via
ACTION_VIEW; failures surface as a non-blocking snackbar via openError.
* fix: remove redundant out-of-order io.mockk.* wildcard import (ktlint CI)
ktlint import-ordering requires lexicographic order; the trailing wildcard
'import io.mockk.*' sorts before 'io.mockk.coEvery' and was redundant — every
mockk symbol is already explicitly imported, and the single mockkConstructor
call uses a fully-qualified io.mockk. prefix (needs no import).
Fixes ktlint failure on PR #725.
* fix(ci): mock android.jar Intent stub so openAttachment GATEWAY test passes
android.jar unit-test stubs Intent.setDataAndType/addFlags to throw
'... not mocked'. openWithView builds an Intent and calls them, so
openBytes' runCatching swallowed the exception and startActivity never
fired, failing the test. Mock the Intent constructor + setters and
capture the constructed instance to assert the ACTION_VIEW contract.
* fix(#724): resolve host media download URLs and deduplicate REST synced chat bubbles
* fix(#724): add FLAG_ACTIVITY_NEW_TASK for file intents and update GatewayFileClientTest
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* feat: gateway files client — fetch any gateway-hosted file over HTTP
Add GatewayFileClient: a typed client for the gateway's managed-files
download endpoint (GET /api/files/download?path=<enc>&token=<enc>), which
streams the raw bytes of any file living on the gateway host. This is the
mobile equivalent of desktop's mediaExternalUrl() and works on a remote
phone (real HTTP, not a host-local file read).
- buildDownloadUrl(): pure authed-URL builder (null on blank creds /
non-absolute path), mirrors desktop's token-in-query auth.
- fetch(): OkHttp GET via the shared client, maps 401/403/404/413 to typed
GatewayFileResult variants; returns GatewayFile(name, mime, bytes).
- MediaTypes: IMAGE/AUDIO/VIDEO/FILE classification by extension (mirrors
desktop MEDIA_BY_EXT) so future consumers can present images, audio,
video, or generic files — beyond just images.
Pure builders/classifiers are unit-tested. Mobile-only, backend untouched.
This is the foundation PR; issue #724's MEDIA: rendering will consume it.
* fix: decode filename + handle download-URL in mediaNameFromPath
- GatewayFileClient.parseFilename now URL-decodes the captured value so
filename*=UTF-8''a%20b.png yields 'a b.png'.
- MediaTypes.mediaNameFromPath extracts the filename from the percent-encoded
path= query param of a full /api/files/download?path=<enc> URL (the real
case), not just a bare path.
* fix(remote): fix unit tests for gateway file client and media parsing
Replace plus-encoded spaces with %20 in download URL builder, URL-decode filenames in parseFilename, and extract target filename from URL path parameters.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Assistant messages could not display inline images or attachments on the
mobile client. The markdown renderer had no image block, and
AssistantBubble only rendered MarkdownText + timestamp, ignoring
message.attachments (only UserBubble did).
- Add MdBlock.Image + standalone  parsing in MarkdownText.
Accepts http(s) gateway URLs and data:image/...;base64 inline media.
- Render message.attachments inside AssistantBubble (mirrors UserBubble)
via InlineAttachment / AsyncImage.
- Add parser unit tests for image blocks (url, data: URL, inline-in-paragraph).
Local verification: ./gradlew ktlintCheck compileDebugKotlin assembleDebug
all green; app-debug.apk produced.
Fixes feature gap where agent-delivered media (e.g. MEDIA:/ images
sent to chat) was invisible in the Android app.
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* Adopt /api/health REST route (issue #713)
Cheap liveness probe on the connection screen: GET /api/health surfaces
backend version + auth mode (Token vs Gated/OAuth) without the heavier
/api/status payload. Added HermesApiService.getHealth + HealthStatus model;
ConnectViewModel.loadHealth probes on open (best-effort, silent on failure);
ConnectScreen shows a read-only health card when reachable.
Verified: ktlint clean, assembleDebug BUILD SUCCESSFUL locally.
* Add used/full context meter to chat screen
Shows session prompt tokens vs model context window above the composer,
polled every 5s. Color-coded bar warns at 70%/90%.
* fix(chat): fix unit tests for context usage meter
Format >=100k tokens cleanly in ContextUsageChip and stub AuthManager profile ID in ChatViewModelTest.
* Fix context meter: source used tokens from input_tokens
Gateway /api/sessions/{id} has no last_prompt_tokens column; use
input_tokens instead. Show meter as soon as full context is known.
* Add tappable context breakdown sheet
Chip opens a bottom sheet with input/output/cache/read/write/reasoning
token rows + message count. Source breakdown from /api/sessions/{id}.
* Refresh context meter on streaming complete
Trigger fetchContextUsage immediately when a message finishes streaming
instead of waiting for the 5s session-sync poll.
* Fix context meter: source used tokens from live WS session.info
The REST /api/sessions/{id} endpoint reads the DB input_tokens column,
which is stale/0 for active sessions — that's why the meter was stuck at 0.
Switch the meter to the live values the gateway pushes over the WS
session.info payload after each turn:
- used = last_prompt_tokens (real used context for the turn)
- full = context_length (model-accurate window; keep /api/model/info as a
fallback before the first turn completes)
fetchContextUsage now only tops up the denominator from /api/model/info and
no longer overwrites the live WS value with the stale REST input_tokens.
The 5s poll remains as a fallback.
* Revert "Fix context meter: source used tokens from live WS session.info"
This reverts commit 932393fd80993d57e66fe93552a949460d646674.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* Add session.redirect affordance (issue #710)
Route send through session.redirect while a turn is streaming and the prompt
is plain text (no attachments, not a /command); otherwise prompt.submit.
Adds WsMethods.SESSION_REDIRECT + HermesWsClient.sendRedirect. No separate
button — the composer already sends mid-turn, so send just does the redirect.
* fix(chat): capture typing state before sendMessage for redirect
Capture wasStreaming state before updating uiState.isAgentTyping so normal sends call prompt.submit.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
Flush throttled reasoning before state-transition events (MessageStart/
Complete/Done/ToolStart) so a reasoning delta arriving just before a
transition is not dropped. Also compare reasoningText in sameMessages so
a reasoning-only restore actually re-merges.
- ChatStreamingController.flushPendingReasoning() forces a synchronous flush
- handleWsEvent calls it for the transition events
- sameMessages now includes reasoningText in its equality check
- tests: ChatStreamingControllerTest (flush + no-resurrect), reasoning survives
MessageDone after an unflushed delta
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* perf: optimize synchronous I/O in AuthLoginViewModel
Replace blocking OkHttp .execute() with a suspending .await() extension
(CallExt.kt) backed by suspendCancellableCoroutine, freeing Dispatchers.IO
threads during network waits. Probe/connect paths made suspend.
- Call.await() guards both onResponse/onFailure with continuation.isActive
to avoid AlreadyResume on cancellation.
- Removes non-asserting AuthLoginViewModelBenchmark timing dump.
* perf: optimize synchronous I/O in AuthLoginViewModel
Replaced the blocking `.execute()` OkHttp calls with a suspending `.await()` extension function inside `AuthLoginViewModel`. This frees up the `Dispatchers.IO` threads by deferring the blocking operations to OkHttp's internal task runner, providing a safer async approach and preventing Thread pool starvation on slow network responses.
Modified `probeDashboardInternal` and `connectBasicAuth` to be proper `suspend` functions and resolved corresponding unit tests.
* fix(test): add missing import io.mockk.any in AuthLoginViewModelTest
* fix(test): fix mockk imports and Log matchers in AuthLoginViewModelTest
* fix(test): use persistent mockProbeClient in AuthLoginViewModelTest
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
* test: add comprehensive tests for loadSettings in SettingsViewModel
Adds `testLoadSettings_loadsAllProperties` to `SettingsViewModelTest.kt` to thoroughly verify that `loadSettings()` correctly maps all required settings properties from `AuthManager` to the `uiState`. The test utilizes `runTest` and `advanceUntilIdle()` to properly handle testing the coroutine launch in the ViewModel's init block.
* test: add comprehensive tests for loadSettings in SettingsViewModel
Adds `testLoadSettings_loadsAllProperties` to `SettingsViewModelTest.kt` to thoroughly verify that `loadSettings()` correctly maps all required settings properties from `AuthManager` to the `uiState`. The test utilizes `runTest` and `advanceUntilIdle()` to properly handle testing the coroutine launch in the ViewModel's init block. Fixes ktlint formatting issues on imports.
* perf: Hoist regexes in MarkdownText.kt to avoid recompilation
Resolves a performance issue where `bulletRe`, `taskRe`, and `orderedRe`
were instantiated and compiled inside the `parseBlocks` function on every
call, causing unnecessary overhead during repeated markdown parsing (e.g.,
when chat messages stream in).
These regex objects have been hoisted to file-level `private val`
constants (`BULLET_RE`, `TASK_RE`, `ORDERED_RE`), ensuring they are compiled
only once when the class is loaded.
A benchmark `MarkdownTextBenchmark.kt` was added to verify the
performance change, showing an approximate 50ms improvement over 10k
iterations.
* perf: Hoist regexes in MarkdownText.kt to avoid recompilation
Resolves a performance issue where `bulletRe`, `taskRe`, and `orderedRe`
were instantiated and compiled inside the `parseBlocks` function on every
call, causing unnecessary overhead during repeated markdown parsing (e.g.,
when chat messages stream in).
These regex objects have been hoisted to file-level `private val`
constants (`BULLET_RE`, `TASK_RE`, `ORDERED_RE`), ensuring they are compiled
only once when the class is loaded.
A benchmark `MarkdownTextBenchmark.kt` was added to verify the
performance change, showing an approximate 50ms improvement over 10k
iterations.
* perf: Hoist regexes in MarkdownText.kt to avoid recompilation
Resolves a performance issue where `bulletRe`, `taskRe`, and `orderedRe`
were instantiated and compiled inside the `parseBlocks` function on every
call, causing unnecessary overhead during repeated markdown parsing (e.g.,
when chat messages stream in).
These regex objects have been hoisted to file-level `private val`
constants (`BULLET_RE`, `TASK_RE`, `ORDERED_RE`), ensuring they are compiled
only once when the class is loaded.
A benchmark `MarkdownTextBenchmark.kt` was added to verify the
performance change, showing an approximate 50ms improvement over 10k
iterations.
* test: add unit tests for ChatPersistenceRepository
Added ChatPersistenceRepositoryTest to verify the repository
correctly delegates to the ChatMessageDao and maps between
ChatMessage UI models and Room entity objects. Uses the
existing FakeChatMessageDao for reliable in-memory testing.
* test: fix ktlint issues in ChatPersistenceRepositoryTest
Formatted ChatPersistenceRepositoryTest to comply with standard
ktlint rules for the project.
* test: enhance ChatPersistenceRepositoryTest coverage
Added role and content assertions to `persistMessages_upsertsAllToDao`
and added a new `loadMessages_filtersBySessionId` test to verify
session isolation per PR feedback.
* fix(#688): route expired auth sessions to sign-in
Adopt global 401 latch, WS disconnect, and Back/profile guards.
* chore: clean up formatting noise
* fix(ktlint): reorder imports to place java and kotlinx at end
* fix(ktlint): place java.io.IOException at end of imports
* test: fix NavBackStack type parameter in AuthenticationRoutingTest
* test: add AuthSessionState tests to AuthLoginViewModelTest
* fix(ktlint): fix import order in AuthLoginViewModelTest
* fix(#686): implement robust history pagination
Honor server-returned offset, degrade gracefully on older servers, and parse JSON tool results.
* fix(ci): repair unit-test compilation and pagination test mocks
- Remove trailing blank line in SessionMessagesResponse.kt (ktlint)
- Replace unresolved io.mockk.eq/coAnswers usage in ChatViewModelTest with
an ordered-call counter inside coAnswers, which compiles and runs on
MockK 1.14.11 (eq is a matcher-scope member, not a top-level import)
- Align HermesApiServiceMockWebServerTest with the PR's JsonElement content
model (contentText) and the new include_compacted=true default query param
All unit tests pass locally with the ByteBuddy agent preload.
---------
Co-authored-by: M57 <hy4ri@users.noreply.github.com>
When the initial REST page returns empty messages, hasOlderMessages
was still set to true (offset > 0), causing an infinite loop of empty
load-more requests. Guard with && chatMessages.isNotEmpty().
Co-authored-by: M57 <Hy4ri@users.noreply.github.com>
* [feature] issue #634: HTTPS + reverse-proxy server endpoint support
- Add ServerEndpoint model (scheme-aware base URL, resolve/webSocketUrl/relativeRequestPath, cleartext policy)
- AuthManager: baseUrl is canonical; setHost/setPort derive from endpoint; endpointForBuild() for scheme-aware URLs
- ApiClient/CookieManager/HermesWsClient: route through ServerEndpoint (no hardcoded http://)
- Data migration: ServerUrlMigration converts legacy host/port to baseUrl
- UI: Connect/AuthLogin/Settings now use a single Server URL field with transport warning
- Build: ALLOW_CLEARTEXT buildConfigField + manifest placeholder for dev
- Tests: ServerEndpointTest + ServerUrlMigrationTest; adapt existing suite to baseUrl API
Co-Authored-By: M57 <Hy4ri@users.noreply.github.com>
* fix(authlogin): basic-auth derives token automatically, no manual entry
connectBasicAuth() built a fresh OkHttpClient for the password-login call and
a different one for the ws-ticket call, so the Set-Cookie from
/auth/password-login (the session token lives in the cookie, not the JSON body)
never reached /api/auth/ws-ticket. That 401'd the ticket request, returned null,
and the UI fell back to asking the user to paste a token.
Both calls now use the shared OkHttpProvider.probe client (same CookieManager
jar), so the session cookie flows through and the WS ticket mints automatically —
matching the dashboard contract (routes.py: password-login sets Set-Cookie,
ws-ticket requires it).
Add AuthLoginBasicAuthFlowTest proving the cookie propagates on the shared
client (and that ws-ticket 401s without it).
Fixes the basic-auth token prompt reported against issue #634.
* chore: remove TASK-634-TESTS.md (stray task-tracking doc, not part of feature)
* ci: broaden .gitguardian.yml test-file ignores (unit + instrumented + variant dirs)
* feat(authlogin): smart auth-mode detection from /api/status + drop pairing-string login
- Replace 302-sniff probe with authoritative GET /api/status read
(auth_required + auth_providers) to derive DashboardAuthMode.
- Add OAUTH mode (stubbed, 'coming soon' UI, TODO tracks issue #639).
- Render login UI by detected mode: token (loopback), username+password
(basic/gated), or OAuth notice.
- Remove legacy pairing-string box + onPairingString/applyLegacyPairing.
- Extract internal deriveAuthMode() + add 5 unit tests.
Closes nothing; OAuth tracked in #639.
* fix(chat): don't block WS connect on empty token in gated (ticket) mode
connectWebSocket() gated on getToken() != null, correct for loopback
(token) mode but wrong for gated (ticket) mode: the ticket is minted
fresh by HermesWsClient.refreshWsTicketIfNeeded() from the persisted
session cookie, so getToken() is expected empty at connect time. The
early return left the WS unopened -> DISCONNECTED -> chat showed
'reconnect' immediately after basic-auth login.
Only require a non-blank token in non-gated mode.
Fixes#640.
* test: remove onPairingString test cases for deleted pairing-login path
The pairing-string login (onPairingString/applyLegacyPairing) was removed
from ConnectViewModel, so the 5 tests that exercised it (3 in
ConnectViewModelTest, 2 in E2eIntegrationTest) no longer compile. Remove
them. Also drop the now-orphaned connect_pairing_string from values-ko.
Fixes the Unit & Integration Tests CI failure on PR #638.
---------
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>
* 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(#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>
* 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>
- SlashCommandDispatcherTest: covers all branch routes (Interrupt /
NewSession / RpcDispatch) incl. case-insensitivity, args, slash-only,
empty, and unknown commands.
- SlashCommandDispatchRpcTest: asserts a non-hardcoded command is
forwarded to COMMAND_DISPATCH with the exact {name,arg,session_id}
shape the gateway expects, args split correctly, and that client-
handled /stop + /new do NOT hit the backend.
Verified against the live hermes-agent gateway: commands.catalog returns
185 catalog entries but command.dispatch only handles a subset, so most
forwarded commands currently 4018. These tests lock the client contract;
the backend gap is tracked separately.
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>
