The streaming splitter treated every "." as a sentence boundary, so the
operational reply "17.5°C" was read as "17." / "5°C" and "1.8 km/h" as
"1." / "8 km/h" - numbers spoken digit-by-digit plus extra TTS calls.
An ASCII terminator (. ! ?) now only ends a sentence when it is followed by
whitespace, a closing quote/bracket, or end of text. In-token dots (decimals
"17.5", versions "v2.0", hosts "example.com") are followed by a digit/letter,
so they no longer split. CJK fullwidth terminators stay unconditional since
those scripts use no trailing space. Language-agnostic, punctuation only.
- bridge: lookahead-gated boundary regex + finditer-based chunking
- tests: regression cases for decimals (17.5/1.8), versions, URLs, and an
integer that genuinely ends a sentence
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The /converse turn synthesised the entire reply before any audio played, so
time-to-first-audio grew with reply length. Add a streaming /converse_stream
endpoint that emits the transcript/reply first, then one audio clip per
sentence as each finishes synthesising. The Discord voice layer enqueues each
clip on arrival via the existing FIFO playQueue, so the first sentence starts
speaking while the rest are still being synthesised.
STT and the reply engine still run to completion before the first clip; only
TTS is pipelined. The non-streaming /converse and /text endpoints are
unchanged.
- bridge: language-agnostic sentence splitter (bridge/text_utils.py) + NDJSON
streaming route
- bot: ndjson() reader + converseStream() client; voice.ts plays clips
progressively
- tests: splitter unit tests + bot ndjson/converseStream tests
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Empirical A/B/C measurement against the live RTX 5050 Ollama stack
(qwen2.5:3b + nomic-embed-text) showed keep_alive=0 unloads the embed
model ~2s after every call, so each turn after a brief idle gap pays a
cold reload. VRAM is not the constraint (~4.4-4.7 GB free with both
models resident) and keep_alive=0 never evicted the chat model, so CPU
embedding (num_gpu=0) gave no benefit. A short positive keep_alive is
the fastest of the three: it keeps the ~0.3 GB embed model resident
across consecutive turns at negligible VRAM cost.
Add tests/test_embeddings.py covering the warm-across-turns behaviour.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The "디스코드 토큰은 마지막에" section still told users to fill DISCORD_BOT_TOKEN
and call the `/자비스` slash command, but the default mode is the userbot
(selfbot) — a normal bot account cannot Go Live. Rewrite the setup to lead with
DISCORD_SELFBOT_TOKEN + DISCORD_VOICE_CHANNEL_ID and the text-command control
(`!자비스 join`/`leave`), with the burner-account ToS warning, and keep the
normal-bot path documented as the optional legacy alternative.
The bot, bridge and melo worker boot together, but the MeloTTS model takes
tens of seconds to load. If the bot logged in and auto-joined the voice channel
before the voice was warm, the first reply synthesised to nothing and was
silently dropped.
- bridge /health now reports `tts_ready`. For MeloTTS this pings the worker,
which only binds its HTTP port AFTER the model is loaded (main() warms before
serve_forever()), so a successful ping is a precise "voice is warm" signal.
- The bot polls /health and waits for `tts_ready` before logging in. It does
not wait on brain_ready (the reply engine / Whisper load lazily on the first
turn — a slow first turn is fine, a silent one is the bug). After a 180s cap
it proceeds anyway so a TTS load failure degrades to text-only.
Live-verified: startup logs show "⏳ MeloTTS 준비 대기 중" then
"✓ 음성(MeloTTS) 준비 완료 — 로그인 진행" then "✓ 유저봇 로그인", in that order.
Open-Meteo's geocoder only matches Latin/English spellings, so a Korean city
name like "서울" returns zero results even though the place exists. With
OUTPUT_LANGUAGE locked to Korean the tool-calling model naturally fills
`location` with the Korean name, which dead-ended on every weather request
("could not find location").
When the first geocode is empty and the name is non-ASCII, ask the warm small
model for the common English exonym ("서울" -> "Seoul") and retry once. ASCII
names skip the round-trip entirely.
Live-verified: "서울 날씨 알려줘" now returns real Seoul weather. Tests cover the
romanise-and-retry path and the ASCII short-circuit.
The config template never set intent_judge_model, so it fell back to the code
default gemma4:e2b. That model is not pulled by this stack (Ollama only has
qwen2.5:3b, qwen3:8b, nomic-embed-text), so every auxiliary small-model call —
intent judge, tool router, weather place extraction, query decomposition —
targeted a non-existent model, silently failed, and fell open. This crippled
tool routing and argument extraction on the 3B brain.
Render intent_judge_model from a new OLLAMA_INTENT_MODEL env var that defaults
to OLLAMA_CHAT_MODEL, so the auxiliary calls reuse the already-warm chat model
(one resident model, no extra load). tool_router_model="" then resolves through
the chain to the same model.
Verified: rendered jarvis.json now has intent_judge_model=qwen2.5:3b, and the
weather place extractor returns "서울" / "Tokyo" (it returned None for
everything while pointed at the missing gemma4:e2b).
qwen2.5:3b emits tool calls in text shapes the parser dropped, breaking
two reviewer-reported behaviours:
- `getWeather: {"location": "Seoul"}` (a JSON object after the colon) was
dumped wholesale into {"query": "{...}"}, so `location` never reached the
tool. getWeather then ran with empty args, returned the auto-detected
location's weather, the model noticed the mismatch and retried — looping up
to 8 times before giving up with an English error. Now the JSON object after
the colon is parsed directly as the argument dict.
- `call_stop: {"id":..., "function": {"name": "setBroadcast",
"arguments": "{\"action\": \"stop\"}"}}` — a single tool_call object without
the `tool_calls: [...]` array wrapper, behind a `call_xxx:` label — matched
no form, so the raw JSON leaked to the user AND setBroadcast never ran
("방송 꺼줘" did nothing). Now name + arguments are pulled from the embedded
`function` object when the name is in the allow-list.
Field-captured from the live qwen2.5:3b brain (2026-06-12). Tests cover both
shapes, non-ASCII args, dict/string arguments, and unknown-tool rejection.
Harden the reply-language lock so qwen2.5:3b reliably stays in the locked
language instead of leaking the query language back in:
- reply_language_directive(): single resolver with clear precedence —
explicit OUTPUT_LANGUAGE lock wins over the Piper/Chatterbox English-only
fallback (this deployment's actual TTS is Korean MeloTTS, so the legacy
English lock was both wrong and contradicting the Korean lock).
- Stronger, override-explicit directive wording, inserted near the FRONT of
the system prompt so a small model gives it primacy over the persona.
- build_system_prompt(output_language=...): rewrite the persona's "in the
user's language" clause to the locked language so the persona stops
fighting the lock.
- docs/llm_contexts.md: document the resolver, precedence, and placement.
Live-verified on the running brain (qwen2.5:3b): Korean voice-style input
and a cold English query both return fully Korean replies with no CJK/Hanja
leak. Tests cover unset/set/agnostic/whitespace + precedence + persona rewrite.
Add an optional OUTPUT_LANGUAGE env var that forces every reply into a
single language. When set, output_language_directive() injects a "respond
only in <language>" instruction (also forbidding other scripts) into the
chat loop's system prompt, next to the existing TTS English-only lock.
Empty (default) keeps the multilingual "reply in the user's language"
behaviour, so upstream is unaffected.
For the Korean-only deployment this also suppresses the occasional trailing
CJK/Hanja fragment qwen2.5:3b leaks on free-form chit-chat.
- system_prompt.py: language-agnostic output_language_directive() helper
- engine.py: read OUTPUT_LANGUAGE, append directive in _build_initial_system_message
- docker-compose.yml + .env.example: document/pass the new var
- docs/llm_contexts.md: note the new gating on the main reply context
- tests: cover unset/set/agnostic/whitespace cases
Replace the blunt global OLLAMA_KEEP_ALIVE=-1 (which kept every model,
including nomic-embed, resident in VRAM forever) with per-request residency:
- llm.py: all three /api/chat payloads send keep_alive=30m so the actively
used chat model stays resident and voice turns never pay a cold reload.
- embeddings.py: /api/embeddings sends keep_alive=0 so nomic-embed unloads
right after each call instead of squatting in VRAM next to the chat model.
- docker-compose.yml: drop the global OLLAMA_KEEP_ALIVE=-1; document the
per-request scheme on the ollama service.
Switch the default chat model qwen3:8b -> qwen2.5:3b. Verified live on the
RTX 5050 (8GB):
- ollama ps: qwen2.5:3b 2.4GB, 100% GPU (8B was 92% GPU / 8% CPU), UNTIL ~30m
(the 30m pin, not "Forever"); nomic-embed absent after several enriched turns.
- nvidia-smi: ~3.2GB VRAM used total (qwen 2.4GB + whisper 0.7GB) vs ~6.6GB.
- Korean /text turns: warm 1.7-4s (cold first load ~52s), vs ~5-7s on 8B;
time/weather/places tool calls fire and reply in Korean.
Known limitation: qwen2.5:3b can occasionally leak a trailing CJK phrase on
free-form chit-chat (factual/tool replies stay clean).
Default ollama keep_alive is 5 min, so any voice turn after a short idle paid
a full Qwen3 8B reload into the GPU (~30-60s) before replying. Set
OLLAMA_KEEP_ALIVE=-1 so the chat + embed models stay loaded.
Measured: post-idle turn ~60s -> steady-state ~4.5s per turn (model pinned
"Forever" in `ollama ps`).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a dedicated MeloTTS Korean voice (speed 1.5) as the primary TTS engine,
served by a long-lived in-container worker so each Discord turn pays only
inference cost, not model-load cost.
- bridge/melo_worker.py: tiny HTTP service in its own /opt/melo py3.11 venv,
keeps the KR model warm, returns PCM16 WAV on POST /synth.
- bridge/server.py: synthesize() routes to the melo worker first; Piper stays
as an opt-in fallback (MELO_FALLBACK_PIPER, default off so Korean is never
mangled through the English voice). /health reports tts_engine.
- docker/setup-melo.sh: builds the isolated venv (pinned torch 2.12.0 /
torchaudio 2.11.0 CPU, MeloTTS pinned to a commit for reproducible rebuilds),
pre-fetches mecab-ko, and warms a dedicated HF cache (/opt/melo-cache) with a
real KR synth so all BERT + KR checkpoint assets are baked into the image.
- docker/supervisord.conf: runs melo-worker before the bridge with
HF_HOME=/opt/melo-cache (the whisper_cache volume shadows the default HF
cache) plus HF_HUB_OFFLINE/TRANSFORMERS_OFFLINE so it reads the baked cache
and never retries the network on load.
- Dockerfile/.env.example: wire the melo build layer and config knobs.
Verified: offline synth passes with --network none and the prod volume mounted;
prod container recreated, all supervisord services up, bot logged in, and an
end-to-end /tts call returns a 44.1kHz mono PCM16 WAV.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The broadcast subtitle rule treated any ko* track as Korean, so YouTube's
auto-generated (자동 생성) Korean track would switch captions on. Match only
human-authored tracks (kind !== 'asr', vss_id not 'a.*') and pass the full
track object to setOption so YouTube selects the manual track rather than the
same-languageCode ASR one. Captions stay off when only an auto track exists.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The selfbot library's native voice connection cannot complete Discord's current
DAVE/v8 voice handshake and times out (15s). Root cause: DAVE E2EE (mandated
~March 2026). The bundled @discordjs/voice 0.18.0 also lacked DAVE.
Fix: upgrade @discordjs/voice to 0.19.2 + add @snazzah/davey (the DAVE protocol
lib) and drive the selfbot client through @discordjs/voice's joinVoiceChannel /
receiver via the client's voiceAdapterCreator — i.e. reuse the normal-bot
VoiceSession with a user-account channel. Verified: the userbot now reaches a
READY voice connection (host test) and joins the channel in-container with no
timeout. This also fixes the normal-bot voice path under DAVE. Drops the
hand-rolled selfbot-native receiver (audio.ts).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
A user account is the only kind Discord lets Go Live, so add a userbot run mode:
when DISCORD_BOT_TOKEN is absent but DISCORD_SELFBOT_TOKEN is set, the app runs
as a userbot instead of the normal slash-command bot.
Stage 1 (conversation): log in with discord.js-selfbot-v13, auto-join
DISCORD_VOICE_CHANNEL_ID on startup (or "!자비스 join" when unset), listen via the
selfbot VoiceReceiver (pcm), transcribe+reply through the bridge, and speak the
reply back with playAudio. Shared PCM/WAV helpers extracted to audio.ts.
run-bot.sh now starts in either userbot or normal mode. Go-Live broadcast +
broadcast-coupled routing land in stage 2 on the same voice connection.
ToS note: selfbots violate Discord ToS; burner account only.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Addresses review findings on the dockerized stack:
- Container Chrome search was dead: add --remote-debugging-port + a non-default
--user-data-dir (Chrome 136+ refuses CDP on the default profile), add the
playwright dep (browse-search.mjs connectOverCDP) with browser download
skipped, and connect to 127.0.0.1 not "localhost" (container localhost -> ::1
while Chrome binds IPv4). Verified: browse-search returns real results.
- Broadcast toggle reliability: always offer setBroadcast in screen-share mode
(the embedding/keyword router dropped it for non-English utterances) and make
its description force a tool call. "방송 꺼줘"->stop now 5/5; no false triggers.
- Stop the broadcast on voice leave (no orphaned stream).
- Security: bind VNC/noVNC to loopback by default (VNC_BIND override) and the
bridge to the container loopback (BRIDGE_HOST=127.0.0.1), not published.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Completes the STREAM_BROWSER=true behaviour:
- handleJoin auto-starts the broadcast on voice join and wires the session to
the guild streamer; each turn reports the live state to the brain so search
routes Chrome (live) vs Gemini (off).
- New setBroadcast tool lets the user toggle the broadcast by voice ("방송
켜줘/꺼줘") via the LLM (no hardcoded phrases); it refuses when
STREAM_BROWSER=false. The directive flows brain -> bridge (broadcast_action)
-> bot streamer.start/stop, guarded by isActive() so it's idempotent.
- Per-turn IPC uses a thread-local (reply/turn_state.py) instead of threading
params through the whole engine chain: bridge sets broadcasting in, tool
records the directive out; Tool.execute exposes broadcasting on ToolContext.
Bot typecheck clean; brain covered by tests/test_set_broadcast.py (+ existing
routing tests). Specs + docs updated.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Makes the all-in-one image actually run the new real-time-search features and
closes review gaps:
- Gemini OAuth path: install Node 22 + @google/gemini-cli (pinned 0.46.0) in the
image; mount a DEDICATED host dir (~/.config/javis/gemini) holding only the
OAuth creds to /root/.gemini (not the whole ~/.gemini). Verified in-container:
`gemini -p ... -o json` returns a grounded answer with no API key.
- Broadcast audio: add PulseAudio + a headless null-sink (run-pulse.sh, new
supervisor program); export XDG_RUNTIME_DIR/PULSE_SERVER so Chrome playback
and the selfbot `ffmpeg -f pulse -i @DEFAULT_MONITOR@` share one daemon.
Verified: default sink virtual_speaker, monitor present, ffmpeg capture OK.
- Bind the brain bridge to 127.0.0.1 only (internal, unauthenticated API).
- VNC host port is overridable; this server pins VNC_PORT=5902 (.env) since the
host already runs Xvnc on 5901.
Verified in-container with CDI GPU passthrough: RTX 5050 visible, NVENC
encoders (h264/hevc/av1) available.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
STREAM_BROWSER becomes the broadcast *capability* master flag; the live
screen-share state (new ToolContext.broadcasting, passed per turn by the bot)
decides the backend:
- master off -> broadcast disabled, always Gemini
- master on + live on -> on-screen Chrome (visible on the stream)
- master on + live off -> Gemini
context.broadcasting is None outside the voice path (evals, text entry, older
bot) and falls back to the master flag, so current behaviour is unchanged.
This is the brain-side foundation; bot-side wiring (bridge passes broadcast
state, auto-broadcast on voice join, voice on/off toggle) follows.
Specs + docs/llm_contexts.md updated. Covered by
tests/test_web_search_broadcast_routing.py.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
yolo auto-approves every tool call, so a real-time search query could in
principle trigger write/shell tools. Default approval mode still auto-runs the
CLI's read-only web search in headless mode but never silently approves
destructive tools. Verified end-to-end: a grounded query returns a current
answer in ~23s with the account OAuth login. Test asserts yolo is absent;
specs and docs/llm_contexts.md updated.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds a GEMINI_AUTH=oauth (default) sub-mode that shells out to the Gemini CLI
using the user's Google-account login instead of an API key. gemini_cli_search()
runs `gemini -p <query> -o json --skip-trust --approval-mode yolo`, strips
GEMINI_API_KEY/GOOGLE_API_KEY and sets GOOGLE_GENAI_USE_GCA=true so the CLI
selects the account OAuth method and fails fast when no login exists. Bounded by
a 30s timeout and fail-open to the DDG/Brave/Wikipedia cascade on any failure
(CLI missing, login expired, quota 429, timeout). GEMINI_AUTH=apikey keeps the
legacy REST path. Specs and docs/llm_contexts.md updated; behaviour covered by
tests/test_realtime_gemini_cli.py.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Completes the two info modes in the Python brain:
- config.py: read STREAM_BROWSER / GEMINI_API_KEY / GEMINI_MODEL from env into
Settings (stream_browser, gemini_api_key, gemini_model). Verified load_settings
reads both modes.
- realtime_search.py: two fail-open backends returning the same fenced
UNTRUSTED-WEB-EXTRACT envelope: browser_search() shells the Node CDP helper to
drive the on-screen Chrome (visible on the broadcast); gemini_search() calls
the Gemini REST API with google_search grounding.
- web_search.run(): routes by mode before the DDG cascade (true->browser,
false->Gemini), falling through to DDG/Brave/Wikipedia on any miss.
- browse_and_play tool: plays a YouTube video on the shared screen (true mode
only); registered in the tool registry.
- specs + docs/llm_contexts.md updated (new Gemini LLM context); CLAUDE.md spec
registry updated.
Verified live against the running Chrome: true-mode webSearch returned real
Google results for "오늘 서울 날씨", browseAndPlay played the IU 밤편지 MV, and
false-mode degrades gracefully on a bad/absent key. A valid GEMINI_API_KEY is
still needed to confirm the real Gemini grounding output.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
First increment of the STREAM_BROWSER real-time-info modes (true = browser,
false = Gemini):
- browse-search.mjs: drives the on-screen Chrome via CDP so the action shows on
the broadcast. `search` returns the top Google results (title/url/snippet);
`youtube` plays the first result. Verified live: real-time Seoul weather
results, and IU 'Good Day' MV playback.
- .env.example: GEMINI_API_KEY / GEMINI_MODEL for the false-mode Gemini account.
- docs/stream_browser_modes.md: architecture + integration map (brain config,
the two mode-gated tools, registry, design decisions) for the remaining wiring.
The Python brain wiring (config.py mode/gemini fields, browseAndSearch +
geminiSearch tools, registry, specs, llm_contexts) lands next - it needs a
running brain and a Gemini key to verify, rather than committing untested edits
into the 39k-line engine.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Addresses review of the STREAM_BROWSER / broadcast-defaults work:
- SelfbotStreamer now spawns broadcast-helper.mjs on stream start and kills it on
stop/self-end (alongside capture + keepalive). The ad-skip, subtitle rule and
fullscreen-toolbar-hide are therefore guaranteed broadcast-wide defaults tied
to the broadcast - not a manual process. Fail-open: if node/Chrome deps are
absent the stream runs without the helper. Verified the helper is a child of
the broadcast holder and armed.
- Enforce STREAM_BROWSER at the streamer (start() returns early when
screenBrowser===false), so EVERY caller including stream-hold.ts is voice-only
when it's off, not just the slash command. stream-hold.ts reads STREAM_BROWSER.
- Fix broadcast-helper fullscreen: resolve the window of the tab actually in
HTML5 fullscreen (via its CDP targetId) instead of the first HTTP tab, so the
right Chrome window is toggled when multiple windows exist.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Add STREAM_BROWSER (.env) gating screen-share/browser mode. false => the
/자비스 stream command stays voice + API/MCP only (no Go-Live); true (default)
=> screen share as before. (Browser-driven info retrieval in true mode is a
follow-up build; the bot has no browser-control tools yet.)
- Make the two test-time fixes broadcast-wide defaults via broadcast-helper.mjs:
it now also watches every tab for HTML5 fullscreen and toggles Chrome window
fullscreen so the address bar is hidden for ANY video (xfwm4 won't hide it on
'f' alone), restoring on exit. Subtitles were already enforced per video.
scenario.mjs drops its own fullscreen toggle and relies on the helper.
- Revert the test-settings env vars from .env.example (not wanted).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Addresses review of the ad/subtitle work (the ad-skip.mjs -> broadcast-helper.mjs
rename's other half; the prior commit only recorded the deletion):
- ad mute leak: the ad-skipper muted during an ad but never un-muted, so the
main video stayed silent after the first ad. Save the pre-ad muted/playbackRate
and restore them when the ad ends (verified: muted false -> true -> false).
- captions were only applied once when scenario.mjs ran, not for the whole
broadcast. The persistent helper now applies the rule (OFF by default, Korean
ON if offered) per video and ENFORCES it every tick - one-shot did not hold
because YouTube silently re-enabled captions (verified it stays off across 8s).
- ad-skip + captions merged into broadcast-helper.mjs (one CDP process).
- the 60fps MV test now lives in the repo: scenario.mjs gains MV_QUERY (search +
auto-pick the first >=60fps result) and WATCH_SECONDS, plus the
fullscreen-toolbar-hide fix. The broadcast runs via the committed
stream-hold.ts (audio + keepalive), not an out-of-repo copy.
- document the test env vars (CDP_PORT, HOLD_MS, TEST_*, MV_QUERY, WATCH_SECONDS).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Addresses review of the ad/subtitle work:
- ad mute leak: the ad-skipper muted during an ad but never un-muted, so the
main video stayed silent after the first ad. Save the pre-ad muted/playbackRate
and restore them when the ad ends (verified: muted false -> true -> false).
- captions were only applied once when scenario.mjs ran, not for the whole
broadcast. Move the rule (OFF by default, Korean ON if offered) into the
persistent helper so it runs per video, and ENFORCE it every tick - one-shot
did not hold because YouTube silently re-enabled captions (verified it now
stays off across 8s).
- merge ad-skip.mjs + captions into broadcast-helper.mjs (one CDP process).
- the actual 60fps MV test now lives in the repo: scenario.mjs gains MV_QUERY
(search + auto-pick the first >=60fps result) and WATCH_SECONDS, with the
fullscreen-toolbar-hide fix. The broadcast runs via the committed
stream-hold.ts (audio + keepalive), not an out-of-repo copy.
- document the test env vars (CDP_PORT, HOLD_MS, TEST_*, MV_QUERY, WATCH_SECONDS)
in .env.example.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Adds ad-skip.mjs: connects over CDP and injects a watcher into every tab
(current and future) that clicks "Skip ad" the moment it appears, closes overlay
ads, and fast-forwards unskippable ads (seek-to-end + 16x + mute) so they clear
in ~1s. Self-contained (no extension, no hosts/network changes) and reconnects
across Chrome restarts. Documented in the README.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two broadcast-experience improvements:
- Audio: the Go-Live stream was video-only. Capture the desktop sound (the
default PipeWire/Pulse sink monitor, @DEFAULT_MONITOR@) as a second ffmpeg
input and mux AAC into the mpegts; the library re-encodes it to Opus for
Discord. Controlled by STREAM_AUDIO / STREAM_AUDIO_SOURCE (default on). ffmpeg
inherits XDG_RUNTIME_DIR to reach the pulse socket. Verified: the streamer now
reports "Found audio stream" and the monitor carries Chrome audio (~-11 dB).
- Subtitles: in the browse scenario, default captions OFF, but auto-enable a
Korean track when the video offers one (getOption captions tracklist ->
setOption / unloadModule).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
On the streamed VNC desktop (xfwm4), Chrome did not hide its toolbar when a
video entered HTML5 fullscreen via 'f' - the window was full-screen (outerHeight
1080) but the tab/address bar stayed, leaving only 988px of content, so the
address bar bled into the Go-Live broadcast.
Toggle Chrome-initiated browser fullscreen via CDP (Browser.setWindowBounds
windowState fullscreen) around the 'f' step. That reliably hides the toolbar
(innerHeight 1080 vs 988); the toolbar is restored on exit, so normal browsing
still shows it. Verified live: clean full-screen video, no toolbar.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The Go-Live broadcast looked badly choppy: video and scrolling stuttered while
the cursor stayed smooth. Root cause is TigerVNC: it only refreshes its
framebuffer while a VNC client is attached, but the broadcast reads that
framebuffer with x11grab (not as a VNC client). With no viewer attached the
captured screen idled at ~1.5 fps (measured 3/30 distinct frames); the cursor
looked smooth only because x11grab overlays the live cursor on every frame.
- Add a headless RFB keepalive (vnc-keepalive.ts) that stays connected for the
life of the stream and requests incremental framebuffer updates at the stream
framerate. SelfbotStreamer starts it on broadcast start and tears it down on
stop/self-end. Measured 3/30 -> 57/60 distinct frames at 60 fps. Fail-open;
authenticates with VNC_PASSWORD or the ~/.config/tigervnc/passwd file.
- Fix a resource leak: when the Go-Live ended on its own, only the active flag
was cleared, leaving the x11grab->nvenc ffmpeg running forever (pinning a CPU
core while no media was transmitted, with only the gateway TCP left and no UDP
media). The self-end path now tears down capture, keepalive and voice like
stop() does.
- Tests for both paths (self-end teardown; keepalive DES auth, port mapping,
password resolution). Add @types/bun so bun:test typechecks; document the
keepalive and recommended Chrome flags in README and .env.example.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
bringIntoView returned the last boundingBox() unconditionally after the
scroll loop exhausted, so an element still outside the viewport would be
clicked anyway. Validate the final box against the actual viewport bounds
on both axes (innerWidth/innerHeight) and return null otherwise, so
humanClick fails instead of clicking an off-screen coordinate.
Address review accuracy: humanClick used DOM scrollIntoViewIfNeeded and fell
back to Playwright locator.click() when an element had no box - neither is real
input. Now it brings elements into view with a real wheel scroll and throws if
there is no on-screen box (no synthetic click). Header comment and README
corrected: xdotool injects synthetic X input (not a physical HID device), and
all actions are real input while the CDP/DOM API is used only to read state.
Make every action real keyboard/mouse via xdotool, not just the visible
browsing: address-bar navigation (Ctrl+L + char-by-char typing), the YouTube
settings gear -> 화질 -> 1080p menu (real clicks, verified hd1080), the autoplay
toggle, the play button, and fullscreen via the real 'f' key (F11 isn't honored
by this WM; 'f' yields true 1080p fullscreen without pausing). CDP/DOM API is
now used only to read state for verification.
The startup catch cleared this.active unconditionally. In a stop()+restart
race during the slow login/pauses, the first attempt's catch would fire after
the second start() had already taken the lock, unlocking it mid-startup and
letting a third start() race in. Guard the active/state reset with
`this.controller === controller`, matching the field-null and playStream
.finally guards.
Verified live: stop during login then restart keeps the restart's lock
(active stays true), and it clears to false only once truly stopped; no crash.
The human-pause delays leave start() in-flight for several seconds, which
exposed two races:
- stop() during a pause only ended the pause; start() continued and called
joinVoice on the streamer stop() had already nulled (null deref).
- `active` was set only just before go-live, so a second /stream during the
delay passed the guard and both calls raced on the same overwritten streamer.
Now start() locks `active` before any await, keeps controller/streamer/capture
as local refs, and calls signal.throwIfAborted() after each await so an
interleaved stop() unwinds into a catch that tears down via the local refs and
clears instance state only if it still points at this attempt. isActive() now
reflects "starting" during the delay too.
Verified live: concurrent start is rejected ("이미 송출 중입니다"), stop() mid-
startup returns a cancel message with isActive=false and no uncaught error, and
the happy path still goes live and tears down cleanly. tsc --noEmit passes.
Joining voice and starting the broadcast instantly looks like a bot. Add
randomised, human-plausible pauses (~0.9-2.2s after coming online before
joining the channel, ~2.5-5s after joining before hitting Go Live) so the
cadence isn't machine-instant or fingerprintable. The pause resolves
immediately on stop() so teardown never hangs mid-wait.
Verified live: end-to-end join -> settle -> Go Live took ~8s before the
stream went live, held for 15s, and tore down cleanly. tsc --noEmit passes.
Address review: the capture ffmpeg had no -b:v, so it encoded at nvenc's
low default (~2.47 Mbps) and the library then re-encoded to 8 Mbps, which
only upscaled already-lost detail. The double encode also kept CPU decode
+ scale + re-encode in the library, contradicting the "GPU handles it"
claim.
Now the system ffmpeg produces the final Discord-ready H264 in one pass
(-b:v/-maxrate at the configured bitrate, -bf 0, 1s keyframes, yuv420p,
-forced-idr) and prepareStream uses noTranscoding:true to remux only. One
GPU encode, no library decode/scale/re-encode.
Verified locally: high-motion source fills 8.7 Mbps at these args (vs the
~2.47 Mbps no-bitrate default), real :1 desktop holds 60fps at realtime,
and the capture -> copy/remux chain yields h264 1920x1080 yuv420p 60fps
has_b_frames=0. tsc --noEmit passes. Live Discord test pending reboot.
Bump the default broadcast to 1080p 60fps at 8 Mbps and route both encode
stages through the GPU (RTX 5050, h264_nvenc) so 60fps stays smooth without
loading the 4-core host.
- selfbot.ts: capture ffmpeg uses h264_nvenc when streamHw is on (falls back
to software x264 otherwise), and prepareStream now passes Encoders.nvenc()
so the library's transcode runs on the GPU too. Guard loadLib for Encoders.
- config.ts: VNC_FRAMERATE default 30 -> 60, VNC_BITRATE_KBPS 4000 -> 8000.
- .env.example: document the new 1080p60/8 Mbps defaults and STREAM_HW.
Verified locally: h264_nvenc x11grab holds a steady 60fps with headroom,
Encoders.nvenc() returns valid h264_nvenc settings, and tsc --noEmit passes.
Live Discord voice-channel verification pending a host reboot.
End-to-end verified with a real burner token + voice channel: login OK, posts
to the text channel, joins voice, and Go-Live streams the host :1 desktop.
- selfbot.ts now captures the X display with the SYSTEM ffmpeg (reliable
x11grab) and pipes it into prepareStream, instead of relying on the lib's
bundled libav input devices (not portable). Capture process is killed on stop.
- package.json: trustedDependencies (node-av, @lng2004/node-datachannel) so the
native streaming deps build automatically on bun install (incl. Docker).
- Dropped the unused nvenc path (the lib's exported `nvenc` is undefined at
runtime); software H264 encode for now.
get-token.ts now writes the Remote Auth URL as a 512x512 QR image
(/tmp/javis_qr.png, override via QR_OUT) in addition to printing the link, so
it can be sent to the user and scanned from a second screen with the Discord
mobile app. Adds the qrcode dependency.
bot/src/get-token.ts uses discord.js-selfbot-v13 DiscordAuthWebsocket: it
prints the Discord Remote Auth URL (https://discord.com/ra/<code> — the same
thing a login QR encodes). Open it on a phone with the Discord app, approve the
"New login" prompt, and the user token is written to .env as
DISCORD_SELFBOT_TOKEN. Works from a single mobile device (no second screen, no
password, no browser devtools). `bun run token`.
- voice.ts: reply playback is now a FIFO queue (AudioPlayerStatus.Idle drains
it) so concurrent speakers no longer cut each other's replies off.
- selfbot.ts: rewritten against the REAL @dank074/discord-video-stream v6 API
(verified from its d.ts): prepareStream(input, opts, signal)->{command,output},
playStream(output, streamer, {type:"go-live"}, signal), Streamer.joinVoice.
x11grab via customInputOptions; optional NVENC encode (RTX 5050) via exported
`nvenc`. package.json pinned to ^6.0.0 (was a wrong ^4.2.1).
- Dockerfile: dropped the hardcoded python3.12 LD_LIBRARY_PATH. faster-whisper
>=1.1 self-locates the pip CUDA libs; ldconfig (full path, glob) registers
them as a robust fallback. Verified: ld.so cache lists libcublas/libcudnn and
GPU whisper works with LD_LIBRARY_PATH empty.
- bridge: STT resample 48k->16k upgraded from nearest-neighbor to linear
(np.interp).
Verified: tsc clean, image builds, GPU whisper OK via ldconfig, compose valid.
Code review of the bridge/bot/docker work found:
- TTS bug: bridge called PiperVoice.synthesize(text, wav) but that method
returns AudioChunks and takes a SynthesisConfig as its 2nd arg, not a wav
file -> TTS would fail. Switched to synthesize_wav(text, wav_file).
Verified: produces a valid 22050Hz mono WAV.
- run-bot.sh now waits if ANY of DISCORD_BOT_TOKEN/APP_ID/GUILD_ID is missing
(config.ts throws on a missing one), preventing a supervisor crash-loop.
Verified clean: discord.js Events.ClientReady == 'clientReady' (existing
handler correct); image rebuilds.
GPU acceleration is now on by default and verified end-to-end on the
Blackwell RTX 5050 (sm_120):
- Ollama offloads 100% to GPU (log: library=CUDA compute=12.0,
BLACKWELL_NATIVE_FP4=1). compose passes GPU via CDI
(devices: nvidia.com/gpu=all) to both ollama and javis.
- Whisper STT on GPU: faster-whisper>=1.1.0 + nvidia-cublas/cudnn cu12,
LD_LIBRARY_PATH baked into the image. Verified float16 transcribe on
sm_120; bridge auto-falls back to CPU when no GPU is present.
- Model: default chat model -> qwen3:8b (best 8GB-VRAM tool-calling,
~5GB Q4). Embed stays nomic-embed-text.
- README documents the host one-time setup (nvidia-container-toolkit +
`nvidia-ctk cdi generate`) and GPU on/off.
Verified: image builds; GPU visible in both containers via compose;
ollama ps = 100% GPU; faster-whisper cuda OK + CPU fallback OK;
bridge /health 200.
