Compare commits
16
Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
41037a3897 | ||
|
|
50c5fd8373 | ||
|
|
788d2abcb5 | ||
|
|
3ec432cd8b | ||
|
|
ef5bae7ca5 | ||
|
|
9be6422941 | ||
|
|
3d0b090a64 | ||
|
|
f972284dee | ||
|
|
a0bb195d4d | ||
|
|
038a2a472b | ||
|
|
f8141a6a91 | ||
|
|
c83f85745d | ||
|
|
674d2e34a2 | ||
|
|
7531065bdf | ||
|
|
0b922538f0 | ||
|
|
3166139f9e |
+1
-4
@@ -20,10 +20,6 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/), and this
|
||||
- **Desktop CLI: smoother pairing.** The multi-endpoint probe shows per-endpoint progress and latency; a near-expiry session warns before it fails and prints the exact re-pair command; and a bare `ws://host` (no port) defaults to `:8767`.
|
||||
- **Desktop CLI: voice + consent transparency.** `voice` now surfaces enhanced-voice capabilities (Gemini tone tags / persona, xAI speech tags); the desktop-tool consent prompt is clear that it persists per relay and points at `hermes-relay audit`; and computer-use's observe → grant → act flow is documented in `--help`.
|
||||
|
||||
### Fixed
|
||||
|
||||
- **Realtime Agent: brokered Hermes turns no longer fail with `session_not_found`.** When the Realtime Agent reached back to Hermes for context or tool work, it could hand the API Server a session id that belonged to a different session namespace (the gateway/client store), which the API Server rejected. The broker now mints a valid API Server session and retries the turn once when that happens, and reuses an existing API Server session when the id is already valid. It also reads the API Server's current nested `{"session": {"id": …}}` create-session response (previously only the legacy flat shape), so session creation no longer errored with "created a session without an id." Provider-native turns are unaffected.
|
||||
|
||||
## [1.2.1] - 2026-06-21
|
||||
|
||||
### Added
|
||||
@@ -41,6 +37,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/), and this
|
||||
|
||||
- **Voice override applies in Auto mode.** A chosen per-profile/enhanced voice now takes effect when the engine is on Auto with the relay paired — previously only "Relay" mode applied it. Per-profile voice settings are also namespaced by connection.
|
||||
- **Realtime voice "Stop" stops immediately.** Tapping Stop while the agent is speaking now halts realtime playback at once; over-chatty spoken status is throttled; and long background tasks no longer time out the turn (relay keeps the session alive while the task runs).
|
||||
- **Realtime Agent: brokered Hermes turns no longer fail (relay).** When the Realtime Agent reached back to Hermes for context or tool work, a session-namespace mismatch could make the API Server reject the turn with `session_not_found`. The relay now mints or reuses a valid API Server session and retries once, and reads the API Server's current nested create-session response. Provider-native turns are unaffected.
|
||||
- **Hold-to-talk no longer releases on accidental drift.** The mic button holds until the finger genuinely lifts, instead of cancelling when it drifts off the button.
|
||||
- **Voice overlay is readable.** The voice dropdown panel and its status bubbles are opaque (no bleed-through), and the Focus/Overlay/Exit labels no longer wrap to two lines; invalid engine/route combinations are no longer selectable.
|
||||
- **Connection status overlay clears faster.** Resolved (error/warning) connection toasts auto-dismiss within ~5s instead of lingering.
|
||||
|
||||
@@ -1,5 +1,9 @@
|
||||
# Hermes-Relay — Dev Log
|
||||
|
||||
## 2026-06-21 — Released android-v1.2.1
|
||||
|
||||
Cut the Android 1.2.1 release. Version source (`appVersionName 1.2.1` / `appVersionCode 15`) was already on `dev`; release-prep promoted `CHANGELOG.md` `[Unreleased]` → `[1.2.1]` (**Android-only** — the Desktop-CLI entries and the relay `session_not_found` fix stay under `[Unreleased]` for their own `cli-v*`/`plugin-v*` cuts) and rewrote `RELEASE_NOTES.md`, in-app `whats_new.txt`, the Play release notes, and the Play listing copy, all scrubbed for public distribution. Release PR #102 (`dev` → `main`, `--no-ff`) auto-merged on green CI (merge `39cafc2`); `android-v1.2.1` tagged from the `main` tip triggers `release-android.yml` (validate → signed APK/AAB + checksums + GitHub Release; Play Production *draft* when the service-account secret is set, operator clicks Start rollout). Headline 1.2.1 changes: profile lock, in-app changelog, diagnostics detail + Copy/Share/Create-issue, a dismissable update-available nudge, plus voice/realtime fixes (override applies in Auto, realtime Stop halts playback, steadier hold-to-talk, readable overlay, faster connection-overlay dismiss) and a debug-only Developer-options test harness. `RELEASE.md` §2 gained a per-surface CHANGELOG-split clarification.
|
||||
|
||||
## 2026-06-21 — Realtime Agent API Server session handoff (issue #101)
|
||||
|
||||
**Why.** The Realtime Agent's brokered Hermes path (`hermes_run_task`) could fail two ways when reaching back to the API Server. (1) A caller-supplied `chat_session_id` that originated in a different session namespace (the gateway/client session store) was passed straight to `POST /api/sessions/{id}/chat/stream`, which the API Server rejects with `404 session_not_found`. (2) `_create_session()` only read a flat `id`/`session_id`, but the current API Server returns the created session nested under `{"object":"hermes.session","session":{"id":"api_…"}}` — so creation raised "Hermes API created a session without an id."
|
||||
|
||||
+5
-15
@@ -1,25 +1,15 @@
|
||||
# Hermes-Relay-Plugin v__VERSION__
|
||||
|
||||
**Release Date:** June 20, 2026
|
||||
**Since the previous plugin release:** A new, removable **enhancement layer** that lets the relay teach the agent things only the relay knows — starting with sensitive-media classification — plus provider-aware enhanced voice and an isolated, TUI-tuned tmux for relay terminals.
|
||||
**Release Date:** June 22, 2026
|
||||
**Since the previous plugin release:** Reliability fixes for the Realtime Agent voice path — brokered Hermes turns no longer drop with `session_not_found`, and long-running Hermes work no longer times out a live voice session.
|
||||
|
||||
This release adds a clean way for the relay to extend the agent without forking or touching the user's soul/memory. The first use is **sensitive-media classification**: the relay appends a small, auditable system-prompt block teaching the agent to mark private/NSFW media so the paired phone can blur it — with sensitivity staying model-emitted. It's on by default for relay installs (installing the relay is the opt-in), reversible from the dashboard or an env flag, fully visible over a new audit route, and a complete no-op on vanilla upstream. Voice gains provider-aware controls for Gemini and xAI, and relay terminals now run on a dedicated, correctly-configured tmux.
|
||||
This is a focused patch for the relay's Realtime Agent. When a spoken turn reached back into Hermes for context or tool work, a session-namespace mismatch could make the API Server reject the turn, and long background tasks could let the voice session lapse mid-run. Both paths are now resilient. Provider-native voice turns and vanilla upstream (no plugin) are unaffected.
|
||||
|
||||
## What's changed
|
||||
|
||||
### Added
|
||||
- **Relay enhancement layer + agent-context injection.** A reusable, removable layer that injects auditable, fenced blocks into the agent's system prompt at plugin-load. Fail-open at every step (seam absent / block build throws ⇒ base prompt unchanged), config-gated, and a byte-for-byte no-op on vanilla upstream. Built to be retired per-surface as upstream adds a context hook — the same pattern as the bootstrap route shims. See `docs/plans/2026-06-20-relay-enhancement-layer.md`.
|
||||
- **Sensitive-media classification (first block).** Teaches the agent to mark private/NSFW media with the client's spoiler convention so the phone blurs it per the user's setting. **On by default for relay installs**; opt out with `RELAY_AGENT_CONTEXT_ENABLED=0` or the dashboard toggle. Sensitivity stays model-emitted — no relay-side or on-device classifier. No soul/memory is touched.
|
||||
- **`GET /context/injected` audit route.** The relay exposes exactly what it would inject (loopback-open, bearer-gated remotely), so the injection is never hidden — surfaced in the Android chat "What the agent sees" sheet as "Relay context (server-side)".
|
||||
- **Dashboard Agent-context controls.** The Relay management tab gained a master toggle and per-block toggles (labeled experimental / server-side / removable), shown on-by-default for relay installs.
|
||||
- **Provider-aware enhanced voice (Gemini + xAI).** `/voice/synthesize` accepts per-request overrides so a paired client can steer a Gemini voice/model with expressive tone tags, or an xAI voice with expressive speech tags, without changing the server's global voice config.
|
||||
|
||||
### Changed
|
||||
- **Relay terminals run on an isolated, TUI-tuned tmux.** Sessions spawn on a dedicated tmux server/socket with a generated config — `escape-time 0`, truecolor `tmux-256color`, `mouse`/`focus-events` on, `status off` — so editors and full-screen tools behave correctly without touching the user's personal tmux.
|
||||
|
||||
### Fixed
|
||||
- **Relay voice synthesis no longer leaves temporary audio files behind** on the server.
|
||||
- **Clearer voice errors.** Standard voice rejects an over-long recording before uploading and returns a helpful message for audio the server can't read, instead of a generic HTTP error.
|
||||
- **Brokered Hermes turns no longer fail with `session_not_found`.** When the Realtime Agent reached back to Hermes for context or tool work, it could hand the API Server a session id from a different session namespace (the gateway/client store), which the API Server rejected. The broker now mints a valid API Server session and retries the turn once when that happens, reuses an existing API Server session when the id is already valid, and reads the API Server's current nested `{"session": {"id": …}}` create-session response (previously only the legacy flat shape) so session creation no longer errors with "created a session without an id."
|
||||
- **Realtime voice survives long Hermes runs.** A heartbeat now keeps the realtime voice session alive while a long-running Hermes task is in flight, so the turn no longer times out before the work finishes.
|
||||
|
||||
## Install
|
||||
|
||||
|
||||
@@ -320,6 +320,6 @@ dependencies {
|
||||
testImplementation("io.github.takahirom.roborazzi:roborazzi-compose:1.43.1")
|
||||
testImplementation(libs.compose.ui.test.junit4)
|
||||
testImplementation(libs.compose.ui.test.manifest)
|
||||
testImplementation("androidx.test.ext:junit:1.2.1")
|
||||
testImplementation("androidx.test.ext:junit:1.3.0")
|
||||
}
|
||||
|
||||
|
||||
+2
-2
@@ -1,6 +1,6 @@
|
||||
plugins {
|
||||
id("com.android.application") version "9.2.1" apply false
|
||||
id("com.android.library") version "9.2.1" apply false
|
||||
id("org.jetbrains.kotlin.plugin.compose") version "2.3.21" apply false
|
||||
id("org.jetbrains.kotlin.plugin.serialization") version "2.3.21" apply false
|
||||
id("org.jetbrains.kotlin.plugin.compose") version "2.4.0" apply false
|
||||
id("org.jetbrains.kotlin.plugin.serialization") version "2.4.0" apply false
|
||||
}
|
||||
|
||||
@@ -2,30 +2,30 @@
|
||||
appVersionName = "1.2.1"
|
||||
appVersionCode = "15"
|
||||
agp = "9.2.1"
|
||||
kotlin = "2.3.21"
|
||||
compose-bom = "2026.05.01"
|
||||
kotlin = "2.4.0"
|
||||
compose-bom = "2026.06.00"
|
||||
navigation-compose = "2.9.8"
|
||||
okhttp = "5.3.2"
|
||||
okhttp = "5.4.0"
|
||||
kotlinx-serialization = "1.11.0"
|
||||
kotlinx-coroutines = "1.10.2"
|
||||
mockk = "1.14.9"
|
||||
robolectric = "4.14.1"
|
||||
robolectric = "4.16.1"
|
||||
konsist = "0.17.3"
|
||||
security-crypto = "1.1.0"
|
||||
lifecycle = "2.10.0"
|
||||
lifecycle = "2.11.0"
|
||||
activity-compose = "1.13.0"
|
||||
core-ktx = "1.18.0"
|
||||
datastore = "1.2.1"
|
||||
splashscreen = "1.2.0"
|
||||
markdown-renderer = "0.41.0"
|
||||
coil = "3.4.0"
|
||||
markdown-renderer = "0.42.0"
|
||||
coil = "3.5.0"
|
||||
haze = "1.7.2"
|
||||
mlkit-barcode = "17.3.0"
|
||||
camera = "1.6.0"
|
||||
camera = "1.6.1"
|
||||
play-publisher = "4.0.0"
|
||||
media3 = "1.10.1"
|
||||
androidVad = "2.0.10"
|
||||
spatialsdk = "0.12.0"
|
||||
spatialsdk = "0.13.1"
|
||||
play-app-update = "2.1.0"
|
||||
|
||||
[libraries]
|
||||
|
||||
+1
-1
@@ -1,6 +1,6 @@
|
||||
distributionBase=GRADLE_USER_HOME
|
||||
distributionPath=wrapper/dists
|
||||
distributionUrl=https\://services.gradle.org/distributions/gradle-9.5.1-bin.zip
|
||||
distributionUrl=https\://services.gradle.org/distributions/gradle-9.6.0-bin.zip
|
||||
networkTimeout=10000
|
||||
retries=0
|
||||
retryBackOffMs=500
|
||||
|
||||
@@ -20,7 +20,7 @@
|
||||
|
||||
##############################################################################
|
||||
#
|
||||
# Gradle start up script for POSIX generated by Gradle.
|
||||
# gradlew start up script for POSIX generated by Gradle.
|
||||
#
|
||||
# Important for running:
|
||||
#
|
||||
@@ -29,7 +29,7 @@
|
||||
# bash, then to run this script, type that shell name before the whole
|
||||
# command line, like:
|
||||
#
|
||||
# ksh Gradle
|
||||
# ksh gradlew
|
||||
#
|
||||
# Busybox and similar reduced shells will NOT work, because this script
|
||||
# requires all of these POSIX shell features:
|
||||
|
||||
Vendored
+2
-2
@@ -19,7 +19,7 @@
|
||||
@if "%DEBUG%"=="" @echo off
|
||||
@rem ##########################################################################
|
||||
@rem
|
||||
@rem Gradle startup script for Windows
|
||||
@rem gradlew startup script for Windows
|
||||
@rem
|
||||
@rem ##########################################################################
|
||||
|
||||
@@ -72,7 +72,7 @@ echo location of your Java installation. 1>&2
|
||||
|
||||
|
||||
|
||||
@rem Execute Gradle
|
||||
@rem Execute gradlew
|
||||
@rem endlocal doesn't take effect until after the line is parsed and variables are expanded
|
||||
@rem which allows us to clear the local environment before executing the java command
|
||||
endlocal & "%JAVA_EXE%" %DEFAULT_JVM_OPTS% %JAVA_OPTS% %GRADLE_OPTS% "-Dorg.gradle.appname=%APP_BASE_NAME%" -jar "%APP_HOME%\gradle\wrapper\gradle-wrapper.jar" %* & call :exitWithErrorLevel
|
||||
|
||||
@@ -3,7 +3,7 @@
|
||||
"label": "Relay",
|
||||
"description": "Paired devices, bridge activity, media inspection, and remote access for hermes-relay",
|
||||
"icon": "Activity",
|
||||
"version": "1.2.0",
|
||||
"version": "1.2.1",
|
||||
"tab": {
|
||||
"path": "/relay",
|
||||
"position": "after:skills"
|
||||
|
||||
Generated
+2
-2
@@ -1,12 +1,12 @@
|
||||
{
|
||||
"name": "hermes-relay-dashboard",
|
||||
"version": "1.2.0",
|
||||
"version": "1.2.1",
|
||||
"lockfileVersion": 3,
|
||||
"requires": true,
|
||||
"packages": {
|
||||
"": {
|
||||
"name": "hermes-relay-dashboard",
|
||||
"version": "1.2.0",
|
||||
"version": "1.2.1",
|
||||
"devDependencies": {
|
||||
"esbuild": "^0.25.12",
|
||||
"qrcode": "^1.5.4"
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "hermes-relay-dashboard",
|
||||
"version": "1.2.0",
|
||||
"version": "1.2.1",
|
||||
"private": true,
|
||||
"description": "Hermes-Relay dashboard plugin frontend (IIFE bundle). Loaded verbatim by the hermes-agent dashboard via the Plugin SDK global.",
|
||||
"scripts": {
|
||||
|
||||
+1
-1
@@ -1,6 +1,6 @@
|
||||
name: hermes-relay
|
||||
manifest_version: 1
|
||||
version: 1.2.0
|
||||
version: 1.2.1
|
||||
description: "Hermes-Relay plugin for QR pairing, relay sessions, dashboard management, remote desktop/phone tooling, and optional legacy compatibility diagnostics. Standard chat, Manage, and dashboard voice remain vanilla upstream Hermes surfaces."
|
||||
author: Axiom Labs
|
||||
# All three are OPTIONAL — only needed if you use the relay's extra /
|
||||
|
||||
@@ -19,7 +19,7 @@ See ``plugin/relay/server.py`` for the aiohttp server,
|
||||
# CLI releases use desktop/package.json and cli-v* tags. The /health endpoint
|
||||
# reports this plugin version, and stale values make live diagnosis harder than
|
||||
# it should be.
|
||||
__version__ = "1.2.0"
|
||||
__version__ = "1.2.1"
|
||||
|
||||
from .server import create_app, main # noqa: E402 — must come after __version__
|
||||
|
||||
|
||||
+1
-1
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
|
||||
|
||||
[project]
|
||||
name = "hermes-relay"
|
||||
version = "1.2.0"
|
||||
version = "1.2.1"
|
||||
description = "Hermes-Relay plugin — Android device control toolset, QR pairing CLI, and WSS relay server for hermes-agent"
|
||||
requires-python = ">=3.11"
|
||||
dependencies = [
|
||||
|
||||
@@ -7,8 +7,8 @@ pluginManagement {
|
||||
plugins {
|
||||
id("com.android.application") version "9.2.1"
|
||||
id("com.android.library") version "9.2.1"
|
||||
id("org.jetbrains.kotlin.plugin.compose") version "2.3.21"
|
||||
id("org.jetbrains.kotlin.plugin.serialization") version "2.3.21"
|
||||
id("org.jetbrains.kotlin.plugin.compose") version "2.4.0"
|
||||
id("org.jetbrains.kotlin.plugin.serialization") version "2.4.0"
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -17,7 +17,7 @@ plugins {
|
||||
kotlin("jvm")
|
||||
// Compose compiler — version inherited from the root plugins {} block.
|
||||
id("org.jetbrains.kotlin.plugin.compose")
|
||||
id("org.jetbrains.compose") version "1.10.3"
|
||||
id("org.jetbrains.compose") version "1.11.1"
|
||||
}
|
||||
|
||||
kotlin {
|
||||
|
||||
Reference in New Issue
Block a user