Seven more portable Lua packs lifted from common voice-assistant
playbooks (Siri/Alexa/Алиса/Cortana). All sandbox=full, jarvis.http
+ jarvis.system.exec, zero new Rust code. cargo test still 3/3.
news/ — pulls Lenta.ru RSS (en switches to BBC), regex-extracts the
top 5 <title> entries (skipping the channel header), drops them into
the clipboard, then asks Groq for a 2-3 sentence summary at temp 0.4
and speaks it via SAPI. If GROQ_TOKEN is missing it falls back to
just reading the first three headlines.
currency/ — cbr-xml-daily.ru/daily_json.js. Picks USD/EUR/CNY/GBP/
JPY/KZT by keyword in the phrase, computes value / nominal, diffs
against Valute.Previous, says "X стоит N рублей, поднялся/опустился
на M" with ▲/▼ in the notify body.
crypto/ — CoinGecko simple/price with usd,rub and 24hr_change.
Recognises bitcoin / ethereum / solana / dogecoin / the-open-network.
Reports price + direction in plain Russian.
fun/ — joke / fact / quote / compliment, single ask.lua dispatched by
command_id. Per-command system prompt at temp 1.0; seeds with a
random integer so consecutive calls don't repeat.
wiki/ — ru.wikipedia opensearch → REST summary endpoint. If Groq is
available and extract > 350 chars, asks for a 2-3 sentence retell;
otherwise speaks the first 400 chars verbatim. Full extract goes to
clipboard.
power/ — shutdown_pc / restart_pc / sleep_pc / hibernate_pc /
logoff_user / cancel_shutdown. Single act.lua keyed by command_id.
shutdown/restart use shutdown.exe /s|/r /t 30 with a /c message —
30-second grace window so "отмени выключение" actually works.
sleep uses rundll32 powrprof,SetSuspendState, hibernate uses
shutdown.exe /h.
help/ — "что ты умеешь". Reads the parent commands/ dir via
jarvis.fs.list, sorts subdirs (pack names), pairs each with a
hardcoded Russian/English one-liner description, copies the full
list to clipboard and speaks a count + "see clipboard for full list".
Total command packs is now 33 (was 26).
Iron Man icon regenerated from resources/icons/icon.png at sizes
16/24/32/48/64/128/256 via Pillow (make_ico.py committed alongside);
icon.ico is now a clean 83 KB multi-res for sharp display at any
scale instead of whatever placeholder Tauri originally emitted.
dotenvy added to workspace deps. jarvis-app/main.rs and jarvis-gui/
main.rs both walk up from current_exe() looking for dev.env (up to
5 parent levels), so the Desktop shortcut can launch the exe
directly without a wrapper .bat to pre-set GROQ_TOKEN. Falls back
to dotenvy::dotenv() (cwd lookup) if nothing found near the binary.
5 new portable Lua command packs. Bumps total from 21 to 26.
cargo test -p jarvis-core --lib commands::tests still passes 3/3.
reminders/ — set_reminder. Parses "напомни через N (секунд|минут|
часов) <body>" or English equivalents; understands one Russian word
numerals (один, две, пять, десять, пятнадцать, ...) for the count
when speech recognition returns words instead of digits. Defaults to
5 minutes if number/unit missing. Spawns a detached PowerShell that
Start-Sleeps then fires: BurntToast if installed, System.Speech
SAPI (ru-RU voice preference), and a WScript.Shell.Popup as the
guaranteed-visible last resort.
date_query/ — today / tomorrow / yesterday. Single answer.lua keyed
by command_id, computes the offset via PowerShell Get-Date.AddDays()
in ru-RU culture so the weekday/month come out in Russian, then
speaks via SAPI.
voice_type/ — type_text. Strips the trigger, drops the remainder to
clipboard via jarvis.system.clipboard.set, then synthesises Ctrl+V
through user32!keybd_event so it lands in whatever window currently
has focus. Works in any text field (note app, browser, IDE,
Telegram, etc.).
dice/ — coin_flip / roll_dice / random_number. Single roll.lua,
seeds math.random with os.time + jitter, speaks the result. "1-6"
for dice, "1-100" for random.
stopwatch/ — start / check / stop. Uses jarvis.state.get/set
(persisted via jarvis-core's settings DB) to remember the start
timestamp, computes elapsed via jarvis.context.time.timestamp,
formats as "X сек" / "X мин Y сек" / "X ч Y мин Z сек".
All new packs follow the established patterns (sandbox=full, PS-via-
exec helper where needed, USERPROFILE/SystemDrive everywhere, no
hardcoded paths).
GUI: Footer.svelte gets a small inline "Rust" pill next to the
copyright line (red bg #b7411a, white 11px, rounded). Now obvious
at-a-glance which fork's UI you're looking at when both python and
rust are open side-by-side. Python edition mirrors with a console
banner.
Launcher: run.bat at the repo root loads dev.env (skipping #-comments
and blank lines), then `start ""` spawns target\release\jarvis-gui.exe.
The new Desktop shortcut "Jarvis (Rust).lnk" points at this batch
file with resources\icons\icon.ico as its icon. WindowStyle=7 means
the cmd host stays minimized — only the GUI window pops up.
If the user hasn't built yet, the bat prints the cargo build command
and pauses instead of silently doing nothing.
Five more portable Lua command packs. Bumps the total from 16 to 21
packs. cargo test -p jarvis-core --lib commands::tests still 3/3.
translate/ — translate.lua. Picks the target language from triggers
like "переведи на английский" / "translate to russian"; falls back to
English when only "переведи X" is heard. Sends to Groq with a strict
"translation only" system prompt at temp 0.3. Drops the result into
the clipboard, fires a notify, and speaks via PowerShell SAPI in the
matching ISO voice (ru-RU / uk-UA / en-US / de-DE etc.) so a German
translation actually sounds German.
math/ — math.lua. Strips the trigger ("посчитай" / "calculate" /
"сколько будет" / "what is" / ...), sends the remainder to Groq with
temp=0.0 and max_tokens=64, asking ONLY for the numeric result. If
the model replies "нет" (the system-prompt sentinel for non-math),
plays not_found instead of speaking nonsense. Otherwise speaks
"Получилось X" via SAPI.
theme/ — set.lua dispatched by command_id (theme_dark / theme_light).
Flips AppsUseLightTheme + SystemUsesLightTheme under
HKCU\...\Themes\Personalize via Set-ItemProperty. No reboot needed,
Windows shell re-themes immediately.
projects/ — open.lua + list.lua. Reads/creates
%USERPROFILE%\Documents\jarvis-projects.json (a flat
[{"name": ..., "path": ...}] list). open.lua strips the trigger,
exact-matches then substring-matches the project name, opens the path
in VS Code if `code.cmd` is on PATH, otherwise in Explorer. First
call creates a sample config with jarvis-rust / jarvis-python /
dietpi entries pointing under USERPROFILE so it works on any machine.
list.lua just notifies the configured project names.
sound_panel/ — open.lua. Two-liner: rundll32 mmsys.cpl,,1 (Recording
tab). Useful both as a standalone "открой настройки звука" and as
the place to fix the mic-disabled state that prevents jarvis-app
from starting.
Portability: USERPROFILE / SystemDrive / PATH lookup throughout.
Before: clicking "Запустить" in jarvis-gui spawned jarvis-app, which
silently exited with code 1 (or in release builds, looked like the
console window flashed and closed). The pv_recorder library returns
INVALID_ARGUMENT from pv_recorder_init when Windows Core Audio reports
zero capture endpoints (some other app holding the mic exclusively,
or all input devices disabled in mmsys.cpl). User saw no actionable
feedback.
Now: on recorder::init failure jarvis-app calls notify_mic_problem()
which (Windows-only):
- Fires a long-duration Windows toast titled "J.A.R.V.I.S.: микрофон
не найден" with a hint pointing at mmsys.cpl / Recording.
- Spawns "start ms-settings:sound" so the Sound settings page opens
automatically — user can re-enable the mic in two clicks.
Then the original app::close(1, ...) path runs to keep the same exit
behaviour the GUI's get_jarvis_app_stats poller expects.
Cargo.toml: jarvis-app now pulls winrt-notification (already in
workspace.dependencies via jarvis-core) for the toast.
Also incidentally fixed: the release-build C0000139 (entrypoint not
found) loader crash that was showing up before this change. It went
away after the workspace dep was added and the release relink ran.
Most likely the previous release exe had a stale import table from an
earlier partial rebuild; the clean relink resolves it.
Non-Windows builds get a no-op eprintln so the binary still compiles
for Linux/macOS.
Five new portable Lua command packs. Bumps the total from 11 to 16 packs.
All sandbox=full, all dispatched via jarvis.system.exec or jarvis.fs / jarvis.system
APIs, no new Rust code, post_build.py --sync copies them under target/.
cargo test -p jarvis-core --lib commands::tests passes 3/3 against all 16.
window/ — 8 commands, single dispatch.lua keyed by jarvis.context.command_id:
show_desktop / maximize_window / minimize_window / snap_left / snap_right
/ close_window / restore_all / task_view. _window_helper.ps1 P/Invokes
user32!keybd_event for arbitrary combos like "win+d", "alt+f4",
"win+shift+m". Uses one shared script per pack — much less duplication
than the volume/ approach.
clipboard_read/ — read_clipboard. Pulls jarvis.system.clipboard.get(),
caps preview at 400 chars, then shells PowerShell System.Speech to actually
speak it aloud (auto-picks the first ru-RU voice if present). Also fires a
notify with a 120-char snippet.
notes/ — add_note + open_notes. add_note strips the trigger phrase from
the recognized voice, appends "[YYYY-MM-DD HH:MM] body\n" to
%USERPROFILE%\Documents\jarvis-notes.txt via jarvis.fs.append (creates
the file on first call). open_notes opens the file in the default editor
via jarvis.system.open. Trigger list covers RU/EN/UA variants of
"запиши" / "запомни" / "write down" / etc.
process_kill/ — kill_process. Trigger-strips "закрой программу" / "убей
процесс" / etc., then runs the remainder through an alias map (хром→
chrome, телега→telegram, спотифай→spotify, edge/edge/едж→msedge etc.,
20+ entries) before invoking taskkill /F /IM. Notifies success/not-found.
websearch/ — search_google / search_youtube / search_wiki / search_yandex.
Single dispatch.lua reads jarvis.context.command_id to pick the base URL,
URL-encodes the query (Lua %02X gsub, not relying on jarvis.http), opens
the resulting link via jarvis.system.open (system default browser).
Portability check: every helper resolves paths via $env:USERPROFILE /
jarvis.context.command_path / jarvis.system.env — no hardcoded C:\ refs.
GUI front-end does not subscribe to IpcEvent::LlmReply, so the LLM
auto-fallback was effectively silent — user got the OK chime but no
spoken answer. Backlog item resolved.
llm_fallback::handle now also invokes speak_via_sapi() after firing
IpcEvent::LlmReply (both for successful replies and the fallback
error message). The text goes through System.Speech.Synthesis.
SpeechSynthesizer, with a preference pass that picks the first
installed ru-RU voice if present (Microsoft Irina / Pavel etc.) so
Russian replies sound right out of the box, and falls back to the
default English voice otherwise.
PowerShell is invoked fire-and-forget (no Wait, stdout/stderr null)
so the voice loop never blocks. A long answer keeps speaking while
jarvis returns to wake-word listening; the mic may pick up some of
the speech which is the same compromise voices::play_* already makes.
A cleaner version would pause pv_recorder for the duration of the
utterance — that goes in the backlog as part of the eventual
unified IpcEvent::Speak refactor.
Toggle: set JARVIS_LLM_TTS=false to disable (e.g. on a server where
the GUI is the speaker). Non-Windows builds get a no-op stub.
LLM auto-fallback (jarvis-app/src/app.rs + jarvis-core/src/config.rs):
- When neither intent classifier nor levenshtein finds a command match,
route the utterance straight to the LLM instead of just playing the
"not found" sound. Triggers ("скажи X", "answer Y") still work and
take precedence — they short-circuit before command lookup.
- Two new config knobs:
LLM_AUTO_FALLBACK = true — master toggle
LLM_AUTO_FALLBACK_MIN_CHARS = 4 — suppress for very short utterances
so background noise doesn't burn
Groq quota
- Requires GROQ_TOKEN; if absent, behaviour is unchanged (play_not_found).
codegen/ command pack:
- Phrases: "напиши код X", "сгенерируй скрипт Y", "write code Z", etc.
- Strips the trigger from the recognized phrase, sends what remains to
Groq with a strict system prompt ("return ONLY code, no fences, no
commentary") at temperature 0.2.
- Parses the JSON content, unescapes \n / \" / \\ / \t, strips any
remaining ```lang ... ``` fences, drops the result into the clipboard
via jarvis.system.clipboard.set.
- Notifies a 120-char preview + plays ok-sound. Works for any language
the model handles (Python by default if unspecified).
- GROQ_TOKEN / GROQ_MODEL / GROQ_BASE_URL read from env at call time —
same envvars the voice-loop fallback already uses.
ocr/ command pack:
- Phrases: "прочитай экран", "что на экране", "read screen", etc.
- Captures the primary screen via System.Windows.Forms + System.Drawing
to a temp PNG, then shells to tesseract.exe (-l rus+eng for ru/ua,
-l eng for en).
- Resolves Tesseract by PATH first, then C:\Program Files\Tesseract-OCR
and the x86 install dir; if none works the user gets a friendly
"winget install UB-Mannheim.TesseractOCR" hint in a notification.
- Recognized text goes to clipboard + a 200-char preview notification.
All three features are portable (env-var resolution, no hardcoded user
paths). Command-pack total is now 11. cargo test -p jarvis-core --lib
commands::tests passes 3/3.
cargo test on jarvis-core was failing with `LNK1181 cannot open input
file "libvosk.lib"`. The link-search path is declared in jarvis-app's
build.rs (and jarvis-cli's), but the test binary for jarvis-core has
no such config and the workspace .cargo/config.toml didn't cover it.
Adds a minimal build.rs that emits `cargo:rustc-link-search=native=...`
pointing at lib/windows/amd64 (resolved relative to CARGO_MANIFEST_DIR,
no hardcoded absolute paths so the project still builds for anyone
cloning under a different drive/path).
Verified: cargo test -p jarvis-core --lib commands::tests now passes
3/3 (every_command_toml_parses, lua_command_scripts_exist,
every_command_has_phrases).
media/ — play_pause, next, prev, stop via user32!keybd_event (VK_MEDIA_*
0xB0..0xB3). Helper _media_helper.ps1 mirrors the volume/ pack pattern,
so the same C# P/Invoke style works for the four standard media keys
without needing any external tooling. Catches Spotify, foobar, browsers,
Yandex Music PWA — anything that responds to system media keys.
sysinfo/ — battery / time / cpu / ram / disk and an "all" combo.
_sysinfo.ps1 uses CIM queries (Win32_Battery, Win32_OperatingSystem,
Win32_LogicalDisk, Win32_Processor + Win32_PerfFormattedData_PerfOS_
Processor) — no external deps. Time is rendered in ru-RU culture
("01:17, 15 мая, пятница"). Battery falls back to a friendly "no
battery, looks like a desktop" when Win32_Battery is empty. SystemDrive
env var resolves the right disk on non-C:/ installs.
Each sysinfo command notifies the line as a Windows toast and logs it.
Triggered locally: sysinfo_all currently reports:
Сейчас: 01:17, 15 мая, пятница
Батарея не обнаружена (видимо, десктоп)
CPU: AMD Ryzen 7 5700X3D 8-Core Processor, загрузка 4%
RAM: 23.4 из 47.9 ГБ (49% занято)
Диск C:: свободно 118.2 из 1906.8 ГБ
Total command packs is now 9 (verified via jarvis-gui startup log).
Portability: all PowerShell helpers use $env:SystemDrive and CIM
classes that ship with every Windows install; Lua wrappers resolve
the helper path via jarvis.context.command_path (no hardcoded user
profile paths or absolute C:\ refs).
Three unit tests catch the kind of bug that broke weather/set_city (a
phrases array instead of lang→array map silently dropped the whole pack):
- every_command_toml_parses: every resources/commands/*/command.toml round-
trips through toml::from_str::<JCommandsList>. Reports all failures at
once instead of failing on the first.
- lua_command_scripts_exist: for every type="lua" command, the named (or
default script.lua) script file exists in the pack dir.
- every_command_has_phrases: structural commands (terminate/stop_chaining)
excluded, every other command has ≥1 phrase across all languages.
Tests use env!("CARGO_MANIFEST_DIR") -> ../.. -> resources/commands so they
work from any cwd.
NB: cargo test does not run on this machine right now — aws-lc-sys needs
cl.exe (MSVC Build Tools missing). The release binary was built earlier
with MSVC available; toolchain install is a follow-up. The tests are still
valid for CI / a re-armed dev box.
All four packs are pure Lua (sandbox=full) and call out to PowerShell helpers
where COM / native APIs are needed. No new Rust code, no rebuild required:
post_build.py --sync copies them under target/{debug,release}/resources/.
volume/
- volume_up / volume_down: 5x VK_VOLUME_UP|DOWN via keybd_event
- volume_mute: VK_VOLUME_MUTE
- volume_max: 50x VK_VOLUME_UP (system clamps)
Helper _volume_helper.ps1 P/Invokes user32!keybd_event.
apps/
- open_browser : start "" https://www.google.com (uses system default browser)
- open_notepad : notepad.exe
- open_calculator: calc.exe
- open_explorer : explorer.exe
- open_terminal : wt.exe, falls back to powershell.exe
- open_settings : ms-settings: protocol
- open_task_manager: taskmgr.exe
- lock_screen : rundll32 user32.dll,LockWorkStation
- screenshot : System.Drawing capture to ~/Pictures/jarvis_screenshot_*.png
file_search/
- find file <query> : recursive Get-ChildItem in Desktop/Documents/Downloads
(depth 3, top 5 matches). Strips Russian/English/Ukrainian trigger
prefixes from the recognized phrase, then opens explorer /select on the
first hit and notifies with the total count.
output_device/
- list_output_devices : enumerates active render endpoints via
IMMDeviceEnumerator and notifies friendly names with indices
- next_output_device : cycles the system default render device via the
undocumented IPolicyConfig::SetDefaultEndpoint (eConsole/Multimedia/
Communications). Tested locally — listed 7 devices, current correctly
identified at index 5.
set_city used a plain `phrases = [...]` array which fails the lang-map schema
in JCommand — the whole weather pack was being skipped at load. Normalized
to per-language phrases.
browser/command.toml pointed at Priler's bundled AutoHotKey exes that do not
ship in this fork, so the entries were dead. Renamed to .disabled (the loader
skips packs without command.toml) so we do not collide with the new Lua
apps pack while keeping the directory for reference.
Adds webrtc-vad-driven listening window for the post-wake-word phase.
State machine in jarvis-core::audio_processing::vad::listen_window;
adapter in vad::webrtc; integrated in app.rs::recognize_command and
stt::vosk::finalize_speech. Same parameters as Python v0.2.0
(aggressiveness 2, end-silence 1200ms, min-speech 500ms,
min-listen 1000ms, hard-cap 15000ms).
Wires LlmClient into the voice loop with a trigger-prefix pattern
(скажи / ответь / произнеси). Conversation history bounded at 8
turns, system prompt always preserved. LlmReply published via IPC;
pre-recorded play_ok cue accompanies the text reply. On error,
plays play_error and sends russian fallback line over IPC.
mark the v0.3 promise from the v0.2.x section as done and add a new
subsection that documents the GROQ_TOKEN env var, trigger phrases,
the LlmReply ipc event and the rollback-on-error behaviour.
new module crates/jarvis-app/src/llm_fallback.rs holds an optional
LlmClient + ConversationHistory built once at startup. if GROQ_TOKEN is
unset the module logs a warning and stays disabled — voice commands keep
working as before.
both the wake-word voice path (recognize_command in app.rs) and the
gui-side text command path (process_text_command) now check whether the
recognized phrase starts with one of the configured trigger phrases
(ru: 'скажи' / 'ответь' / 'произнеси'). when it does, the remainder of
the phrase is sent to Groq and the reply is published as a new
IpcEvent::LlmReply { text } so the gui can speak it.
on api error the trailing user turn is popped from history, the russian
fallback line is sent over ipc and a 'error' voice cue plays. the loop
itself never panics.
add LLM_DEFAULT_ENABLED / LLM_DEFAULT_MAX_HISTORY / LLM_DEFAULT_MAX_TOKENS
defaults, the russian J.A.R.V.I.S. system prompt, a fallback error line,
and per-language helpers get_llm_trigger_phrases() and get_llm_system_prompt()
so the voice loop can opt into Groq with a 'скажи …' style prefix.
split llm.rs into a module with separate client and history submodules.
ConversationHistory holds an optional system prompt plus a FIFO of user/
assistant turns capped at max_turns; oldest turns evict on overflow,
system prompt is preserved across truncation and clear().
Adds Groq (OpenAI-compatible) LlmClient in jarvis-core behind the
new 'llm' cargo feature, plus a jarvis-cli 'ask <prompt>' subcommand
for one-shot testing. Not yet wired into the wake-word/intent loop —
that's the next feature. thiserror added to workspace deps.
Adds a Russian-language section describing the new jarvis-core::llm
module, the Groq-related environment variables (GROQ_TOKEN,
GROQ_BASE_URL, GROQ_MODEL), the one-shot 'jarvis-cli ask' usage and
a short Rust example. Notes that the module is intentionally not yet
wired into the voice/intent loop — that lands in v0.3.
One-shot invocation: jarvis-cli ask "..." reads GROQ_TOKEN, calls
LlmClient::complete and prints the reply to stdout. Exits 2 when the
env var is missing, 1 on API/HTTP failure. Also available as an 'ask'
command inside the interactive REPL.
New jarvis-core::llm module providing a blocking client for
OpenAI-compatible chat completions endpoints (Groq by default).
- LlmClient::new / from_env (GROQ_TOKEN, GROQ_BASE_URL, GROQ_MODEL)
- complete(messages, max_tokens) -> String
- thiserror-based LlmError / ConfigError
- gated behind a new llm feature, included in default jarvis_app
Not yet wired into the wake-word/intent loop; that lands in v0.3.
Make all 4 workspace crates build cleanly: add build.rs to jarvis-cli
and jarvis-gui for vosk link path, build frontend assets so tauri
generate_context! finds them, gitignore tauri-regenerated schemas.
Workspace feature unification pulls vosk into jarvis-gui through
jarvis-core, so jarvis-gui needs the same rustc-link-search as
jarvis-app and jarvis-cli for libvosk.lib at link time.
The build script now runs 'routify build' before svelte-check so the
generated routes module exists when type-checking. vite build still
runs last to produce dist/client/, which tauri::generate_context!
needs at compile time for jarvis-gui.
desktop-schema.json and windows-schema.json under crates/jarvis-gui/gen/schemas
are regenerated by tauri-build on every compile and only produce diff noise.
