The gaming-PC rerun scored the fast model's honest refusals ('...할 수 있는
도구는 없습니다', '...준비되어 있지 않습니다') as failures, so the kit
downloaded two more fast-model candidates for nothing. Rescored with the
fixed judge, UD-Q4_K_XL, UD-Q6_K_XL and Q8_0 all pass 27/27; the
recommended model's Q2_0/IQ3_XXS/IQ3_S still fail 3/27 by calling an
unrelated tool when no tool fits. Record both results in the v2 docs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add a 'desktop' stage to the M0 kit that installs the pinned agent-desktop
1.2.0 Windows build (SHA-256 checked) and checks, on the gaming PC:
calculator buttons by automation id (12+3=15), set-value and read-back on
a test page in a throwaway browser profile, Discord/Explorer read time,
and a warm MCP server against one CLI process per call. Results keep only
counts, timings, error codes and the calculator digits.
Record agent-desktop (UI Automation) as the planned M7 control path in the
v2 design docs, with raw Windows input only for screens without
accessibility.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The temperature and minute checks dropped the sign and the decimal part,
so "13.9도" and "-13도" passed as 13 degrees. Numbers are now read whole
("-", "영하" and decimals included) and must equal the tool result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The tools stage decides whether to keep a model or measure the next one,
so a made-up answer must never pass. Each case now has a judge for the
spoken answer: numbers must match exactly with their unit (15 or 50
minutes fail a 5-minute timer), weather answers must agree with the rain
in the result and give no other temperature, and a refusal fails when it
also says the thing was done. A new case asks for a timer with no timer
tool offered, the shape Jarvis has today. Unit tests cover the reviewer's
counterexamples and that a lying model makes the stage move on to the
next candidate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
When the skipped-tool guard could not run the planned call (not a literal
call, or arguments the tool does not take) it delivered the answer it had
held back, which was given without the tool and could be a made-up value.
The planned call is now recorded as a failed tool result and the model
answers from that, so the user hears that it could not check or do it.
M0 kit: a case passes only when the final answer uses the tool result
(13 degrees, not a made-up 24) or, for the no-tool request, says it
cannot do it; the no-tool request now counts. The kit mirrors the
engine's failed-result path, and a failing model moves on to the next
candidate on these stricter checks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
User rule: if the tool-calling fix does not work, change the model. The
tools_fast / tools_recommended stages now measure the next, less
compressed model of the same family (fast: UD-Q6_K_XL then Q8_0;
recommended: IQ3_XXS then IQ3_S) when the default still gets a scored
case wrong, stop at the first one that passes every case, and record it
as tools_verdict.chosen. Candidates without enough free disk are skipped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
M0 on Strata caught both v2 models skipping a tool they had been given:
the recommended model made up "24 degrees, sunny" instead of calling the
weather tool, and the fast model said "timer set" without calling the
timer tool (3 of 3 each). qwen2.5:7b does the same through the engine.
- Prompt: TOOL_HONESTY_RULES in the persona (live facts only from a tool
result this turn, actions only claimed after a tool reports success, no
answer text next to a tool call), restated at the end of the system
message by tool_honesty_reminder() with the turn's real tool names.
- Engine: skipped-tool guard. When a native-tool model answers before any
tool has run and the plan names a callable tool as a concrete call, the
answer is dropped and the plan steps are direct-executed, then the model
answers from the results. Only literal, schema-valid steps run (regex
fast path, no LLM repair); otherwise the held-back answer is delivered.
- Planner: the AVAILABLE TOOLS lines now carry argument names
(planner_tool_catalog), so concrete steps use the tool's real keys.
- Eval evals/test_tool_skip_confabulation.py; on qwen2.5:7b, 3 runs each,
available-tool cases went from 0/12 to 9/12 calling the tool.
- M0 kit: tools_fast / tools_recommended stages (run_m0_tools.bat) rerun
tool calling before and after the fix on the real models, loading the
production system_prompt.py and planner.py.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Keep the fast model as the default: the recommended model decodes faster
(131 vs 82 tok/s) but starts every voice answer, long-context reply and
image answer later. Note its working VRAM shrink for game mode and its
fabricated weather answer for the M3 evals.
A named share question now captures the popout straight away when that
person is sharing, even with several sharers (user decision).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Only capture a named person's share when they are the sole streamer or
the user has confirmed the selection in the popout. A timeout no longer
captures whatever is showing, a non-streaming target gets an honest
answer, and clicking the popout after confirmation asks again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Discord allows only one popped-out share window and picks the stream
inside it, so there is never a choice between windows. Use the voice
channel's streaming state instead: one streamer means capture directly,
two or more means asking the user to pick that person in the popout
when the requested person changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The first run left three results out of the plan: the fast model skipped
the timer tool and claimed it was set, the popped-out share window carries
the voice channel name rather than the streamer's, and a covered popout
keeps drawing. The recommended model failed to start (DFlash2 needs a
native pack), so M0 stays open for that one stage with the fixed kit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The user wants the app to work on any Windows PC from its own installer,
so the Parsec driver is a test aid for M0, not a product dependency.
Scope the target to x64 Windows 10/11 outside S mode (ARM64 needs WHQL)
and re-verify the Discord share, cursor and unplug behaviour with the
driver that ships.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The user wants the bot window on no real screen at all. Add a vdisplay
stage that drives the Parsec Virtual Display Adapter already on the PC
(no install, the display dies with the script), moves the clock window
onto it as a normal and as a tool window, and asks about real monitors,
Alt+Tab, Discord's Screens tab and whether the cursor can wander there.
Record the M0 defaults the user confirmed in docs/v2.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Strata 0.1.59 gives Qwen3.8-Flash-Next Q2_0 the DFlash2 drafter alone
(--draft-fallback none), but DFlash2 needs a native pack and the canonical
Q2_0 pack is not one, so the engine stops at start ("DFlash2 is off here:
it needs a native pack ... the start ends here"). Ask for the MTP drafter
instead; Strata falls back to no drafter when the MTP layer is missing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The first gaming-PC run failed the recommended model with only Strata's
one-line reason ("no CUDA-capable device"), which is the picture
reader's side message, not the cause. The engine log stayed on the PC.
A launch that does not become ready now copies the newest
*.engine.log (scrubbed, last 200 KB) into logs/, which the results zip
already packs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The context test re-estimated characters per token from the server's
prompt_tokens after every request. A server that reports a tiny count
made each next prompt grow by orders of magnitude, so the kit tests
reached 8 GB and were killed by the host memory guard. Clamp the
estimate to 0.5-4 characters per token.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- A Strata install or start that never reached a server is saved as
"failed" (not "partial"), so the next run retries the stage.
- A launch that exits non-zero is not ready, even if something answers
on the port afterwards.
- Before the cold start, the stage stops if another program already
listens on port 8080 and tells the user to close it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One double-click kit (scripts/m0/run_m0.bat) that measures the gaming PC
and packs the results into a single zip:
- PC info (GPU, driver, RAM, disks, monitors)
- Windows checks with the user: bot browser window hidden from Alt+Tab
but listed in Discord's share picker and still drawing while covered;
Discord pop-out stream window identification and occlusion; installed
games, anti-cheat services and the borderless game detector
- MeloTTS, CosyVoice2 and Fun-CosyVoice3 Korean synthesis speed
- faster-whisper large-v3-turbo: final/partial latency, CER, wake-word
hint, batched decoding for several speakers
- Strata LLM (fast and recommended presets): TTFT by context length,
decode speed, tool calls, picture reading per quality tier, Korean
answers, parallel requests, VRAM give-back and restart/switch timings
uv and Python, per-stage venvs, models and Strata all live inside the
work folder. Results are scrubbed of user/computer names and home paths.
Verified on Linux GPU (LLM client against Ollama, STT, MeloTTS,
CosyVoice2/3) and on the Windows 11 CI VM (window checks, game scan,
MeloTTS/CosyVoice2/STT installs, Strata install and plan).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Chat channel name/description edits also rename the real Discord
channel and topic by default (toggle kept)
- Memory: separate toggles for registered chat channels and other
speakers in the same voice channel, both on by default
- Screen-share popout is requested by voice, never automated
- v1 head bb60c4d preserved as the remote legacy branch
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Replace the root README with the v2 plan (game PC fully local, Windows
management app) and split the plan into docs/v2/*.md: architecture,
voice-and-llm, conversation, memory, desktop-app, pc-features, privacy,
roadmap, decisions. The v1 README moves to docs/v1/README.md with its
relative links fixed; .env.example points to the new location.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Switch the default TTS engine from online Edge TTS to offline GPU
MeloTTS (privacy-first: reply text never leaves the host). Adds a
supervisord melo-worker program that runs in full/bot/brain roles and
synthesises Korean on the GPU (MELO_DEVICE=cuda, MELO_SPEED=1.5), a
baked /opt/melo py3.11 venv layer in the Dockerfile, and compose/env
defaults. Edge TTS stays available as an opt-in online fallback via
TTS_ENGINE=edge.
Verified on the RTX 5050 (Blackwell sm_120): melo-worker logs
"ready (lang=KR speed=1.5 device=cuda)", /health returns ok, and a
/synth round-trip returns a 177KB WAV.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Run the GPU-heavy compute (Whisper STT, Ollama LLM, TTS) on a dedicated
brain host and the bot + browser on a separate no-GPU app host, talking
over the LAN.
- supervisord: new `brain` role (bridge only) and `app` role (bot +
desktop/Chrome, no local bridge); `full`/`browser`/`bot` unchanged.
- compose: make BRIDGE_URL overridable (was hardcoded to loopback, which
prevented the bot from reaching a remote bridge).
- docs: DEPLOY.md "GPU split" layout with per-host .env blocks; document
the new roles in run-if-role.sh and .env.example.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
For the split where the LLM runs on a separate LAN host (e.g. a GPU box at
192.168.10.9) and the app stack runs elsewhere (192.168.10.5), the brain must
reach Ollama over the network. The javis service hardcoded
OLLAMA_BASE_URL=http://ollama:11434, ignoring any .env value, so the app could
only use the in-stack Ollama. Make it ${OLLAMA_BASE_URL:-http://ollama:11434}:
default unchanged (all-in-one), overridable to point at an external Ollama.
Document the external-Ollama setup in .env.example and DEPLOY.md, refresh the
browser-host IP examples (.9→.5), and de-hardcode the .9 example in novnc.ts.
moveMouse returned ok:true even when humanHover did nothing (no on-screen box)
or the selector never matched — recreating the "claims it moved but didn't"
bug. Now: humanHover returns a boolean (and brings the element into view first);
moveMouse returns ok:false when the target isn't found or has no on-screen box,
and when site=naver/... whose box isn't on the current page it navigates to the
site home first before hovering. search now reports input=human|api-fallback|api
so a silent fallback to cursor-less DOM input is visible, and the tool surfaces
that note in the reply instead of implying a human-like search happened.
The tool had no cursor-move/hover action, so "move the mouse to the search box"
had nothing to call and a weak model just claimed it had moved it. Add a
moveMouse action wired to human.humanHover (real xdotool cursor), targetable by
CSS selector or site= (that site's search box). Also clarify in the tool
description that 'navigate' types into the address bar (no mouse) while
'search'/'type' move the real cursor to the on-page box and click before
typing, so the model picks the visible-cursor path when the user wants it.
detect_model_size only matched 1b/3b/7b, so a genuinely small model such as
qwen*:4b fell through to LARGE and got the terse, less-guided prompt set it
can't follow — contributing to off-tone, rambling replies. Extend the small
patterns to cover all sub-8B sizes (1b-7b) across :/-/_ separators and sync the
spec table.
A 4b model replied to "하이" with "테니스 연습을 Trenches Gym에서..." — it copied
the literal "box at Trenches Gym" few-shot example embedded in the persona
prompt and mangled boxing into tennis, presenting a prompt example as if it
were a real user fact. Remove the copyable proper-noun example and add an
explicit guard: use ONLY names/places/activities that literally appear in the
memory section, never borrow them from the instructions or example wording.
A deployment had OLLAMA_INTENT_MODEL=qwen2.5:7b while the chat model was a 4b,
so every auxiliary call (intent judge, tool router, place extraction, query
decomposition) ran on the bigger, slower model and added latency to each
command. Make the .env.example comment state the invariant explicitly.
Replies stayed long because the prompt stack gave conflicting length signals:
the persona said "one sentence (two at the very most)" AND told the model to
"state the answer in a sentence, then add a dry observation" (a 2nd sentence),
while voice_style said "two to three sentences maximum". The model followed the
longest. Make all three sources agree on exactly one sentence: the persona's
aside must now fold into the same sentence as a trailing clause (never a 2nd
sentence), voice_style caps at one sentence, and agents/llm.md says 한 문장.
Shorter replies also cut Edge-TTS latency, since synth time scales with text
length. Specs (prompts.spec.md) and docs/llm_contexts.md updated; deterministic
prompt-contract tests added.
Whisper was decoding with beam_size=1 (greedy), the least accurate setting,
which hurt recognition on short/accented/noisy Discord-mic speech. Switch the
default to beam search (5, Whisper's own default) and stop conditioning on the
previous clip's transcript (which causes repetition/drift on isolated short
utterances rather than helping). Both are env-tunable (STT_BEAM_SIZE,
STT_CONDITION_ON_PREV) so accuracy/latency can be traded without a code change;
wired into docker-compose and documented in .env.example.
Follow-up to the OLLAMA_INTENT_MODEL split: document that the Docker stack runs
intent judging / tool routing / extraction on a small qwen2.5:3b (pulled by
ollama-init) kept separate from the big chat answer model, and that setting
OLLAMA_INTENT_MODEL = OLLAMA_CHAT_MODEL folds them back onto one resident model.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Intent judging, tool routing and arg extraction are classification/JSON calls,
not the spoken answer, yet the stack aliased OLLAMA_INTENT_MODEL back to the big
chat model — so each command paid the big model's cost 2-3 times for routing
before the reply even ran. With the GPU on, that round-trip stacking is the main
remaining per-turn latency. Default OLLAMA_INTENT_MODEL to qwen2.5:3b (the
project's reference small model, clean Korean on classification) and have
ollama-init pull it. The reply engine already routes these calls through
intent_judge_model, so answer quality is untouched; set OLLAMA_INTENT_MODEL =
OLLAMA_CHAT_MODEL to fold back onto one resident model.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
agents/llm.md promises "play the Nth video from the top", but browseAndPlay
only ever clicked the first result. Add an optional 1-based index argument
(default 1, backward-compatible) threaded to the Node helper, which now clicks
the Nth a#video-title and clamps to the number of results returned so asking
beyond the list plays the last available video instead of failing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Spoken (TTS) replies are 1-2 sentences, so an unbounded num_predict only
exposes the worst case where the chat model rambles or loops. Add an
ollama_num_predict config (default 512, 0 disables) wired into the reply
loop's chat call on both the native- and text-tool paths. The 512-token
headroom stays well above this app's short tool-call JSON, so capping never
truncates a tool call. This keeps the user's quality model instead of
downgrading it. Configurable in the container via OLLAMA_NUM_PREDICT.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- run-chrome.sh: render web content (YouTube/Google/Naver) in Korean. --lang
only sets Chrome's UI; the Accept-Language sent to sites comes from the
profile's intl.accept_languages, which a persisted user-data-dir kept at
en-US. Seed the profile to ko-KR and add --accept-lang=ko-KR,ko.
- agents/llm.md: runtime instructions for the reply LLM (loaded by the agents
feature) — name "자비스", concise 1-2 sentence TTS replies (no emojis/markdown),
nuance-based intent, and YouTube voice controls (open/search/play Nth/pause/
back) via the on-screen browser tool.
- CLAUDE.md: drop the "use emojis in CLI output" rule — this assistant replies by
Discord voice, not CLI, so output should be plain.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
load_settings() coerced any tts_engine outside {piper, chatterbox} to piper, so
with TTS_ENGINE=edge the reply engine saw "piper" and treated the voice as
English-only in reply_language_directive() (only the OUTPUT_LANGUAGE lock kept
replies Korean). Add "edge" (and "melo") to the accepted set so the engine is
labelled multilingual correctly.
Also: a stale tts_engine in the persistent /data/jarvis-settings.json (melo/xtts
from an earlier voice, no longer built) would override the configured engine via
the entrypoint merge and leave the bot silent. Reset those to the env engine
during the merge.
Verified: load_settings() with tts_engine=edge now returns "edge"; the merge
maps melo/xtts -> edge; reply_language_directive("edge") is multilingual; 27
tests pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The user chose Microsoft Edge TTS, voice ko-KR-HyunsuMultilingualNeural at rate
+45% (~1.45x), as the natural Korean voice. Wire it into the bridge and make it
the default engine.
- bridge/server.py: _edge_synthesize() calls edge-tts and transcodes the MP3 to
PCM16 mono WAV with the system ffmpeg (temp file for a correct header);
TTS_ENGINE default -> edge; EDGE_TTS_VOICE / EDGE_TTS_RATE env-driven
- requirements-bridge.txt: add edge-tts (lightweight; httpx)
- compose/.env.example/README: TTS_ENGINE=edge + EDGE_TTS_* knobs; note the
online/privacy trade-off (reply text is sent to Microsoft, needs internet)
- drop the now-unused MeloTTS build layer (Dockerfile) and melo-worker
(supervisord) — edge synthesises in-process, no model/worker baked, slimmer
and faster image; settings UI engine list -> edge/piper, restart only bridge
Verified on host: edge-tts -> ffmpeg yields a valid 16-bit mono 24kHz WAV;
envsubst renders tts_engine=edge; docker build --check + 26 tests pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
docker/jarvis-config.template.json hardcoded "tts_engine": "piper". entrypoint
renders it into /app/config/jarvis.json, and bridge _tts_engine_setting() reads
that JSON BEFORE the env — so TTS_ENGINE=melo in .env was ignored and the bot
synthesised Korean with the English Piper voice (the "foreign accent" the user
heard); the warm melo-worker sat unused.
Template now carries ${TTS_ENGINE}; compose sets TTS_ENGINE=${TTS_ENGINE:-melo}
so envsubst renders the real engine. Verified: envsubst with TTS_ENGINE=melo
yields "tts_engine": "melo", and `docker compose config` passes TTS_ENGINE=melo.
Added a regression test that the template stays env-driven and renders the
configured engine.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
MeloTTS's single Korean speaker sounded non-native ("foreign accent"). Swap it
for Coqui XTTS-v2 with the built-in female studio speaker "Ana Florence"
(language ko), the natural voice used in earlier local runs.
- bridge/xtts_worker.py: new warm HTTP worker (own /opt/xtts venv), same
/synth + /health contract and PCM16 output as the old melo worker
- docker/setup-xtts.sh: builds the venv with cu128 torch (Blackwell) + Coqui
TTS and bakes the XTTS-v2 model offline. Pins transformers>=4.57,<5 (5.x
removed isin_mps_friendly, breaking XTTS) and installs the [codec] extra
(torch>=2.9 needs torchcodec) — both verified by a real host synth
- Dockerfile: replace the melo build layer with the xtts layer
- supervisord.conf: melo-worker -> xtts-worker, env passthrough for
XTTS_DEVICE/SPEAKER/LANGUAGE (always set via compose defaults)
- bridge/server.py: default TTS_ENGINE=xtts, route to the xtts worker, generic
worker-synth helper, neural-only fallback flag (XTTS_FALLBACK_PIPER)
- settings UI: engine dropdown xtts/piper, drop the dead melo_speed field, fix
the supervisorctl restart target to xtts-worker
- compose/.env.example/README: XTTS_* vars, speaker/language knobs, remove melo
- remove bridge/melo_worker.py and docker/setup-melo.sh
- tests: xtts treated as multilingual (not English-only)
Verified on host: coqui-tts loads XTTS-v2 and synthesises Korean as
"Ana Florence" to a 16-bit mono 24kHz WAV.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
supervisord.conf passed MELO_DEVICE through as %(ENV_MELO_DEVICE)s but pinned
MELO_SPEED="1.5", so lowering MELO_SPEED in .env had no effect — the worker
always got 1.5. Pass MELO_SPEED through with %(ENV_MELO_SPEED)s and set a
compose default (MELO_SPEED=${MELO_SPEED:-1.5}, same pattern as MELO_DEVICE) so
the supervisord expansion always resolves and an .env value actually changes
the speaking rate. Default rate is unchanged (1.5). melo_worker logs the
resolved speed at startup, so the env->worker path is verifiable.
Verified: _resolve_speed() returns 1.1 for MELO_SPEED=1.1 (1.5 otherwise), and
`MELO_SPEED=1.1 docker compose config` renders MELO_SPEED: "1.1" into the env.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Drop Markdown files into an agents/ folder and their contents are appended to
the main reply LLM's system prompt, so an operator can extend the assistant's
rules/tone without code changes. Files are concatenated in filename order
(use 00-, 10- prefixes to control ordering) and re-read once per turn, so edits
apply on the next reply with no rebuild/restart. Fail-open: a missing, empty,
or unreadable folder yields no instructions and never breaks a reply.
- load_agent_instructions() in system_prompt.py (AGENTS_DIR env, default
/app/agents); reads *.md only, skips blanks, ignores non-dir paths
- engine.py appends it alongside the existing settings-UI llm_instructions,
under the same "Additional instructions from the operator:" framing
- docker-compose.yml bind-mounts ./agents:/app/agents:ro and sets AGENTS_DIR
- agents/example.md.sample starter template (.sample is not loaded)
- tests cover ordering, md-only filtering, blank-skip, env/arg resolution,
and fail-open paths
- README, .env.example, docs/llm_contexts.md updated
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The bridge only logged Whisper's device on the CPU-fallback path, so a
successful GPU (or silent CPU) load was invisible. Print the CTranslate2-
resolved device on success and on the fallback load, so it is verifiable that
STT is actually running on cuda alongside ollama (GPU) and MeloTTS (cuda).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The Go-Live broadcast encodes with h264_nvenc, but the image only requested
NVIDIA_DRIVER_CAPABILITIES=compute,utility. The NVIDIA Container Toolkit gates
which driver libraries it injects by capability, and the NVENC/NVDEC libs
(libnvidia-encode.so.1 / libnvidia-decode.so.1) come with the `video`
capability. Without it the broadcast ffmpeg dies with
"Cannot load libnvidia-encode.so.1", the capture produces no packets, and
Go-Live never connects, while CUDA workloads (ollama/whisper/melo) and
nvidia-smi keep working because compute+utility are present.
Add `video` so hardware encode is available. Applies to both Linux (CDI) and
Windows Docker Desktop (WSL2).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
.gitattributes pins *.sh to LF, but that only helps after a full working-tree
renormalise, which a Windows build box may not have done. The image build kept
failing at `RUN bash setup-melo.sh` because the checked-out file still had CRLF,
so bash read line 18 as `set -euxo pipefail\r` and aborted with
"set: pipefail: invalid option name".
Strip CR from setup-melo.sh before running it, and normalise all docker/scripts
shell scripts to LF after the app COPY so their shebangs (entrypoint, run-*.sh)
also survive a CRLF checkout. Makes the build EOL-agnostic regardless of host
autocrlf settings.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Windows checkouts with autocrlf=true inject CR into docker/*.sh, so inside
the Linux container `set -euxo pipefail` is read as `pipefail\r` and bash
aborts with "set: pipefail: invalid option name", failing setup-melo.sh and
the whole image build. .gitattributes already pinned .bat/.cmd/.ps1 to CRLF
but never pinned .sh, leaving all nine container scripts exposed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Windows users following the docs hit "The system cannot find the file
specified" because COMPOSE_FILE's separator is OS-specific (':' collides
with the C: drive letter). Fix every Windows example to use ';', add an
explicit OS-separator warning in .env.example, README, DEPLOY.md and the
gpu-windows compose comment, and point users at the explicit `-f` form as
a separator-agnostic alternative.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Make the browser search helper search the way a person does: load the
site home page, type the query into the search box one key at a time, and
press Enter — for both Google `search` and `youtube` — instead of jumping
straight to a results URL. Supports the goal of a human-like assistant.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Add CHROME_USER_DATA_DIR so the browser search fallback can open Chrome
against a dedicated, Google-signed-in profile instead of a fresh anonymous
session. A returning signed-in profile is what actually avoids Google's
/sorry bot-detection page, so this is the reliable way to get browser
Google search in plain text turns. Fallback order is now CDP (broadcast
Chrome) -> persistent profile (when configured) -> ephemeral headless,
all still fail-open to the DDG/Brave/Wikipedia cascade.
Document the profile in .env.example and web_search.spec.md.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>