Two new Settings → AI knobs that compose cleanly with what already
shipped (aiTier, LLM model, translator prompt framing).
**B.1 #15 — Named cleanup presets.** LlmPromptPreset enum
(Default / Email / Notes / Code) appends a short context hint onto
the CLEANUP_PROMPT just before generation. Presets shape tone and
structure ("email paragraph", "bulleted meeting notes", "preserve
technical terms") without licensing the content-editing the
translator-not-editor framing forbids. cleanup_transcript_text_cmd
now takes `preset: Option<String>` which runs through the new
LlmPromptPreset::parse (normalises aliases like "meeting-notes",
collapses unknown values to Default).
**A.1 #28 — Sequential-GPU guard.** New LocalEngine::unload drops
the backend + model_id so a subsequent load actually reclaims VRAM.
load_llm_model, load_model, and load_parakeet_model Tauri commands
grow an optional `concurrent: bool` argument. When concurrent is
Some(false), loading LLM first unloads whisper+parakeet, and vice
versa — prevents VRAM OOM on tight-VRAM setups. Default is the
previous parallel behaviour so nothing changes for multi-GB cards.
Transcribe-in-progress paths (transcribe_pcm, transcribe_file, live)
pass None, so mid-dictation model loads don't accidentally tear
down the LLM.
Settings UI (AI section):
- Cleanup preset segmented button + descriptive copy for each option.
- GPU concurrency segmented button with explicit trade-off text
("faster transitions vs fits in tight VRAM").
Frontend wiring:
- settings.llmPromptPreset flows from DictationPage's
cleanupTranscriptIfEnabled into the Tauri command.
- settings.aiGpuConcurrency flows from both DictationPage (auto-load
on record) and SettingsPage (manual load/unload buttons) as
`concurrent: "parallel" === true` to the load commands.
Tests: three new preset cases in crates/ai-formatting/src/llm_client.rs
(parse aliases, suffix non-empty for non-default, default suffix
empty). All 139 existing lib tests still pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ufal/whisper_streaming #161 documents the classic Whisper streaming
failure: on ambiguous audio the model falls into a prompt loop,
cascading a single token for 10+ words ("I I I I I I I I I I I…").
The chunk-boundary duplicate detector in live.rs doesn't catch
this — the repeat is within a single chunk, and the text is
technically novel so FTS is happy to keep it.
Fold the detection into is_hallucination as a third pass (after
HALLUCINATION_MARKERS substring-match and HALLUCINATION_TRAIL_PHRASES
exact-match). has_consecutive_repetition walks the token stream
(whitespace-split, lowercased) and returns true when any run of
≥REPETITION_RUN_THRESHOLD (4) identical tokens is found.
Threshold chosen deliberately: three consecutive matches appear in
normal speech ("no no no, that's wrong"), four almost never does.
Tests pin both sides — "I I I I I" detected, "no no no" allowed,
alternating patterns ("I am I am I am I am") allowed regardless of
length.
Phrase-level repetition ("thank you thank you thank you thank you")
is a documented companion failure mode but needs a sliding n-gram
matcher — deferred with a code comment flagging it.
No caller changes: post_process_segments already drops
is_hallucination hits when anti_hallucination is enabled.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Whisper was trained on subtitle corpora, so silence and room tone
trigger caption-style artefacts that the previous three-marker
blocklist ("[blank_audio]", "[music]", "[silence]") didn't catch:
"Thanks for watching!", "Please subscribe.", "ご視聴ありがとうござ
いました", "♪♪♪", etc. Documented in WhisperLive #185 / #246 and
ufal/whisper_streaming #121 as the top streaming-transcript-quality
issue after chunk-boundary repeats.
HALLUCINATION_MARKERS widens from 3 to 16 entries: all common
bracketed non-speech tags (applause / laughter / inaudible /
background noise / sounds), parens variants, and musical notation
(♪ / ♫). Still contains-match so the marker triggers even when
Whisper wraps it in other noise.
HALLUCINATION_TRAIL_PHRASES (renamed from AUTO_THANKS_PHRASES) jumps
from 4 to ~30 entries: YouTube sign-offs, subtitle-credit leakage,
and the two most common non-English variants (Japanese "thanks for
watching" + MBC Korean news sign-off). Stays exact-match so
legitimate dialogue containing "thanks" or "subscribe" mid-sentence
never gets dropped — a new regression test pins that invariant.
The <15-char length gate on trail phrases is removed; some of the
new entries (e.g. "please subscribe to our channel.") are longer.
Exact-match against a known list is safety enough.
No caller changes: post_process_segments already drops segments for
which is_hallucination returns true when anti_hallucination is on.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The previous prompt led with "You are a transcript cleanup assistant"
and listed cleanup rules. That framing quietly licenses the LLM to
treat cleanup as content editing — rephrasing for clarity, summarising
long sentences, "improving" phrasing. That's precisely the failure
mode OpenWhispr / Scriberr / Whispering users complain about ("the
LLM changed my meaning").
New framing lifts Whispering's published baseline: "translator from
spoken to written form — not an editor trying to improve the content."
Adds an explicit rule: do NOT improve, summarise, expand, or rephrase;
faithful written-form translation only, never content editing.
Both load-bearing concerns are now regression-tested — the existing
prompt-injection hardening assertions stay, and a new test pins the
translator framing + explicit no-editing rule against drift during
future refactors.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Ports the kon-llm model_manager resume pattern the rest of the way
into kon-transcription::model_manager:
- download() now validates an existing complete file against its
sha256 before skipping; a hash mismatch removes the file and
re-fetches, instead of serving a corrupt file to whisper.cpp.
- download_file() now distinguishes 206 Partial Content, 200 OK
(resume silently ignored by server), and other statuses, rather
than treating any non-206 as 'just use it as a fresh start'.
200-on-Range is handled by discarding the partial and starting
over cleanly.
- New tests: download_file_resumes_from_partial_and_verifies_sha
(TcpListener fixture, same shape as kon-llm's), and
download_file_fails_on_sha_mismatch_and_cleans_part_file.
- sha256_of_file helper + unit test for the existing-file guard.
Dev-deps: tempfile + tokio(net/io-util/macros). Total workspace
lib-test count: 116 → 123.
Co-authored-by: jars <jakejars@users.noreply.github.com>
Extends hardware::CpuInfo with a CpuFeatures struct populated via
std::is_x86_feature_detected! on x86_64 and an architectural
assumption for aarch64 (NEON). Adds has_ggml_baseline() so callers
can cheaply ask 'will whisper.cpp / llama.cpp ship a fast path on
this CPU?' without knowing the arch-specific rule.
The point of #7 is giving the runtime a way to surface a clear
"non-AVX2 fallback" warning before the user hits a wall of silent
slowness. The banner itself ships in the next commit as part of
get_runtime_capabilities (item #1). Tests cover the baseline helper
on both x86 and non-x86 targets.
Co-authored-by: jars <jakejars@users.noreply.github.com>
Adds a build.rs guard that parses Cargo.lock and panics on Windows
if the tokenizers crate ever appears in the workspace dependency
tree, mirroring the MSVC C-runtime conflict that broke Whispering
v7.11.0 when they linked whisper-rs-sys + tokenizers in the same
binary.
On non-Windows hosts the guard downgrades to a cargo:warning so
cross-compilation or CI from Linux surfaces the issue before a
Windows build attempt actually panics.
No tokenizers crate is in the tree today; the guard is preemptive.
If we ever legitimately need HF tokenizers on Windows, the escape
hatch is an out-of-process sidecar (separate CRT).
Co-authored-by: jars <jakejars@users.noreply.github.com>
sqlx 0.8's default feature set pulls in `any`, `macros`, `migrate`, and
`json`. Grepping the workspace confirms none of these are used — the
code calls sqlx::query() / query_scalar() at runtime, implements its own
migration sequencing in crates/storage/src/migrations.rs, is sqlite-only
(no `any` needed), and never derives FromRow / applies sqlx proc-macros.
Dropping them keeps only what's needed: runtime-tokio + sqlite.
Why it matters disproportionately on Windows: the `macros` feature pulls
sqlx-macros → sqlx-macros-core → proc-macro2 / syn / quote / async-trait
/ url / heck / dotenvy / sha2 / filetime. Each proc-macro crate on
Windows MSVC compiles to a .dll with a full linker invocation (slower
ABI than Linux/macOS proc-macro .so). Net: tens of seconds shaved off
every cold-cache CI run, compounding with the cache-path fix in the next
commit.
kon-mcp was already lean (default-features = false); matching that shape
across the workspace now.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
QC smoke sweep flagged two clippy -D warnings lints in code this branch
introduced:
- crates/core/src/process_watch.rs — collapsible_if on the meeting-pattern
match loop, merged the two conditions with &&.
- crates/mcp/src/lib.rs — let-else on the id unwrap that short-circuits a
notification, switched to ? since handle_message already returns Option.
All other clippy lints under -D warnings (audio/capture, hotkey/linux,
storage/file_storage, diagnostics, the duplicate-detection helpers in
live.rs) predate this branch and are out of scope.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
search_transcripts already backs onto the transcripts_fts virtual table
(migration v4, trigger-maintained) via MATCH + ORDER BY rank. Adding a
test to lock the behaviour: token matching is case-insensitive, rank-
ordered, and non-matching tokens return nothing. This is Phase G of the
post-OpenWhispr audit — semantic embeddings stay deferred until the FTS
experience actually hits a wall.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Default off. When on, the layout polls detect_meeting_processes every 15s
with the user's app-name patterns. On a fresh match (edge-triggered — no
re-toast until the app goes away and comes back) we fire a reminder toast
that tells the user which meeting app appeared and their global hotkey. We
never start recording on this signal; the ideology rule says the user
decides. The signal is a single channel: process list match only — no mic
activity heuristic, no calendar.
Backend adds kon_core::process_watch::{list_running_process_names,
match_meeting_patterns} over sysinfo, exposed to the frontend as the
detect_meeting_processes Tauri command.
Settings ships two new fields — meetingAutoCapture (bool) and
meetingAutoCaptureApps (string[]) — with a comma-separated input in the
Output section. Default app list is ["zoom", "teams"], user-editable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
New workspace binary crates/mcp exposes Kon's SQLite store to external
agents (Claude desktop, Cline, any MCP client) without running the Tauri
app. Newline-delimited JSON-RPC 2.0 on stdio, MCP protocol 2024-11-05.
Tools shipped (all read-only):
- list_transcripts — recent transcript summaries, limit 1..200 default 20
- get_transcript — full text + metadata by id
- search_transcripts — FTS5-backed query, limit 1..100 default 20
- list_tasks — all tasks (open + done)
No writes. The Tauri app remains the only writer; kon-mcp just opens the
same SQLite file (via kon_storage::init) and reads. Logs land on stderr to
keep stdout clean for the JSON-RPC stream. Smoke-tested end-to-end with
initialize + tools/list over a pipe.
Wire into an MCP client with:
{ "mcpServers": { "kon": { "command": "/path/to/kon-mcp" } } }
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Parakeet-TDT scores 85 on any GPU-equipped English-capable system (Instant
speed + Great accuracy + GPU boost + headroom) vs ~75 for the best distilled
Whisper. A new test in recommendation.rs locks this in so future scoring
tweaks don't silently regress it.
FirstRunPage previously stored settings.modelSize by title-casing a lowercased
alias — which worked for Tiny/Base/Small/Medium but produced
"Whisper-distil-small-en" for the new distil ids. Swap to an id→label map
and pass the raw model id through to download_model/load_model; the backend
already accepts full ids via the whisper_model_id fallback.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two new registry entries (crates/core/src/model_registry.rs):
- whisper-distil-small-en — 336 MB, ~6× faster than whisper-small-en
- whisper-distil-large-v3 — 1.55 GB, near large-v3 accuracy at medium size
Both are whisper.cpp-compatible GGML binaries hosted on HF by the
distil-whisper org; no runtime change, just wider model choice. English-only
by design (matches upstream Distil-Whisper).
The Settings model picker widens to six options — Tiny, Base, Small,
Distil-S, Medium, Distil-L — ordered roughly by accuracy. Download/load
commands now take the resolved model id (whisper-distil-*) instead of the
lowercased label, so the frontend owns the label↔id mapping.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
kon-llm now owns a real LlamaBackend + LlamaModel, with three Qwen3 tiers
(1.7B Q4, 4B-Instruct-2507 Q4, 14B Q5) selectable per hardware. Downloads
are resumable with SHA-256 verification and stored under ~/.kon/models/llm.
Engine exposes three high-level surfaces — all greedy/temp-0, GBNF-constrained
where output shape matters:
- cleanup_text (prompt-injection-hardened system prompt; profile terms
appended as "preserve these spellings" suffix)
- decompose_task (3–7 micro-steps, constrained JSON array)
- extract_tasks (optional-array; empty when no explicit commitments)
post_process_segments now takes an Option<&LlmEngine> and, when loaded and
format_mode != Raw, joins segments → cleanup → replaces segments with the
cleaned text (first segment span). Rule-based path still runs first; LLM
errors log and keep rule-based output.
Tauri commands: recommend_llm_tier, check_llm_model, download_llm_model,
load_llm_model, unload_llm_model, delete_llm_model, get_llm_status,
cleanup_transcript_text_cmd, extract_tasks_from_transcript_cmd,
decompose_and_store (LLM-backed subtasks).
Settings: AI tier toggle (off / cleanup / tasks), model picker with
downloaded/loaded status, download progress events via
kon:llm-download-progress.
Dictation: ensureLlmModelLoaded on mount, cleanupTranscriptIfEnabled after
stop when tier != off and format_mode != Raw, LLM task extraction when
tier=tasks (regex fallback on failure).
Interim: both llama-cpp-sys-2 and whisper-rs-sys statically link their own
ggml, so src-tauri/build.rs emits -Wl,--allow-multiple-definition on Linux.
Replace with a system-ggml shared-lib setup as a follow-up.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Major quality pass on top of Phase 2. Five substantive changes plus
cross-cutting touches across audio, hotkey, transcription, and Tauri
command layers.
Transcription quality
- Long-audio chunking in commands/transcription.rs: Parakeet and large
file transcription now chunk-and-recompose with overlap trimming, so
the live-path chunking advantage extends to file-based workflows.
- Stateful live speech gate in commands/live.rs on top of the earlier
duplicate-boundary filtering — distinguishes start-of-speech from
mid-speech and holds state across chunks.
Auto-learning corrections
- New crates/ai-formatting/src/correction_learning.rs: extracts user
text corrections from viewer edits and proposes additions to the
active profile's vocabulary.
- src-tauri/src/commands/profiles.rs bridge for frontend-driven
confirmation of learned terms.
- src/routes/viewer/+page.svelte hooks the learning path into the
segment-edit flow so corrections feed profile_terms without a
separate 'train this profile' UX.
Transcript profile provenance
- Migration v8 (crates/storage/src/migrations.rs) adds profile_id to
transcripts, defaulting to DEFAULT_PROFILE_ID so existing rows stay
valid.
- crates/storage/src/database.rs: TranscriptRow + CRUD carry profile_id.
- src-tauri/src/commands/transcripts.rs: add_transcript accepts and
persists profile_id.
- DictationPage.svelte + FilesPage.svelte send activeProfileId on
capture so learned corrections are attributed to the right profile.
Cleanup prompt contract
- crates/ai-formatting/src/llm_client.rs hardened: the CLEANUP_PROMPT
now specifies concrete do/do-not rules, ready for a real model-backed
cleanup pass. The llm_client is still a stub — kon-llm remains unwired
— but the prompt shape is final.
Cross-cutting polish
- Minor touches in audio (capture/decode/resample), hotkey (lib/linux/stub),
core, transcription (concurrency/model_manager/local_engine/whisper_rs),
and the rest of src-tauri/src/commands/*: error-path tightening, log
clarity, TS-migration follow-ups (@ts-nocheck additions for incremental
typing).
Verified locally: npm run check, cargo test -p kon-ai-formatting,
cargo test -p kon-storage, cargo test -p kon --lib commands::live::tests,
cargo check — all green.
Scope boundary: kon-llm crate is still a stub; task extraction remains
rule-based. Bundled local-LLM runtime is the next clean step and is not
in this commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ai-formatting:
- rule_based.rs: collapse_repetitions() merges adjacent duplicate
tokens like 'I I can' -> 'I can' and 'think think that' -> 'think
that'. Normalises case and punctuation before comparison.
- pipeline.rs: post_process now calls collapse_repetitions when
format_mode is Clean or Smart. Added unit coverage.
audio:
- capture.rs: replace the seven deprecated cpal DeviceTrait::name()
call sites with a device_display_name() helper that uses the
non-deprecated description() path. Keeps identical behaviour,
silences compile warnings, ready for cpal upgrade.
Addresses the 'Christ. Christ.' live-transcription boundary duplicate
Jake saw during Group 1 dogfooding. Does not fix all cross-chunk
overlap cases (see live.rs OVERLAP_SAMPLES for the root cause) but
catches the common stutter pattern at post-processing.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
transcribe-rs 0.3.10's ParakeetModel::transcribe_raw ignores its
options argument and calls self.infer(samples, &TimestampGranularity::default())
where default is TimestampGranularity::Token — per-subword segments.
That surfaces in Kon as output like 'T Est Ing One , Two , Three . W Ow ,
This Is T Ri Ble .' because DictationPage joins segment texts with ' '.
Introduce a thin ParakeetWordGranularity wrapper that implements
SpeechModel and overrides transcribe_raw to call the concrete
ParakeetModel::transcribe_with() with ParakeetParams { timestamp_granularity:
Some(Word) }. Pre-existing bug unrelated to Phase 2 work — surfaced during
Group 1 dogfooding because Parakeet was being tested for the first time.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Second dogfood sprint. Headline fix: Linux now uses native KWin/Mutter
decorations instead of fragile frameless `startResizeDragging`, which
collapsed diagonal corner resize to a single axis and made drag feel
laggy. macOS / Windows keep custom chrome via `useCustomChrome` gate.
Other changes:
- Cross-window preferences sync via `kon:preferences-changed` Tauri
event — theme and font changes propagate live to float/viewer.
- Hotkey recorder rewritten to use capture-phase document listener
gated by $effect; button focus was unreliable in webkit2gtk.
- History page redesigned for cognitive-load hygiene: title-first
compact row, inline title input, Edit popout opening /viewer in
edit mode, clipboard export as .md with YAML frontmatter, manual
tag chips + + Add tag input, header tag filter (cap 7), global
Starred filter, `tag:xyz` search syntax.
- `deriveAutoTags` kept as empty hook for post-Task-7 LLM topic tags;
research found all previous auto-tag chips redundant with row UI.
- Viewer window adds edit mode with debounced-save textarea; native
title renamed to "Kon - Transcription Editor".
- Window minimums updated per GNOME HIG + WCAG reflow research:
main 960x600, float 360x480, editor 560x520.
- Microphone picker filters raw ALSA strings (hw:, plughw:, front:,
sysdefault:, null) and dedupes by CARD=X. New `description` field
on DeviceInfo reads /proc/asound/cards so Blue Yeti shows as "Blue
Microphones" instead of the short "Microphones" card name.
- GPU reporting fixed: get_runtime_capabilities now returns
accelerators=[cpu,vulkan] and whisper.supports_gpu=true, matching
the transcribe-rs whisper-vulkan feature linked unconditionally.
- ResizeHandles kept for macOS/Windows frameless: 12px edges, 20px
corners via CSS vars, pointerdown + setPointerCapture, corners
above edges in z-order, rendered as sibling (not child) of the
animated layout root so `position: fixed` is viewport-relative.
- Dueling drag-region handlers removed — `data-tauri-drag-region` and
manual `startDragging()` were stacked on the same elements; kept
the manual handler which has the button/input early-return logic.
See HANDOVER.md for the full session log and deferred items.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
crates/storage/src/migrations.rs:
Replace naive sql.split(';') with split_statements() that tracks
BEGIN...END depth, so migrations containing trigger definitions
execute correctly instead of being split mid-block.
crates/transcription/src/lib.rs:
Re-export transcribe_rs::SpeechModel so callers in src-tauri can
reference it without adding transcribe-rs as a direct dependency.
src-tauri/src/commands/models.rs:
Use Box<dyn SpeechModel + Send> as the load_model_from_disk return
type, matching the trait object that transcription crates produce.
src-tauri/src/commands/transcripts.rs:
Remove the stale .map(|v| v as i32) casts on sample_rate and
audio_channels. InsertTranscriptParams now stores these as i64
(ebf449b), matching the i64 fields on CreateTranscriptRequest;
casting to i32 first would silently truncate large sample rates.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Front-to-back QC pass turned up two independent missing-module
showstoppers (workspace did not compile; frontend did not load) plus a
handful of HANDOVER-claimed features that were wired but dead. Fixes:
P0 — gets the app booting again:
- Add the never-committed src/lib/utils/runtime.js (hasTauriRuntime
detection); 5 imports were resolving to nothing.
- Add the never-committed crates/audio/src/streaming_resample.rs
(rubato-backed StreamingResampler with new/push_samples/flush);
declared in lib.rs and used 3x by live.rs but had no impl.
- Drop the duplicate hasTauriRuntime import in routes/+layout.svelte.
- Allow the transcript-viewer window to use the default capability
(was missing from capabilities/default.json:windows, so the viewer
window could open but not invoke any Tauri command).
P1 — features documented as working but actually dead:
- Pump MicrophoneCapture::take_error_rx() into LiveStatusMessage::
Warning each loop iteration in commands/live.rs. The HANDOVER
promised cpal stream errors would surface as toasts; the channel
was created and never read.
- Replace .expect() on the WebKit media-permission setup with a
logged warning. Failure no longer aborts the whole process.
- Toast on save_preferences failure (preferences.svelte.js had a
silent console.error — now warns once per failure run via the
existing toasts store).
P2 — correctness/robustness:
- add_dictionary_entry: switch INSERT OR IGNORE to ON CONFLICT
DO UPDATE ... RETURNING id so duplicate terms get the real row id
instead of a stale auto-increment.
- search_transcripts: qualify ORDER BY fts.rank.
- InsertTranscriptParams + TranscriptRow: bump sample_rate /
audio_channels from i32 to i64 to match the Tauri DTO and avoid
silent truncation at the boundary.
- Drop the unused tauri-plugin-mcp dependency.
- Promote sqlx in src-tauri/Cargo.toml from linux-only to
unconditional (lib.rs names sqlx::SqlitePool unconditionally —
macOS/Windows builds were latently broken).
- hotkey/linux.rs: stop panicking the hotplug task on inotify
failure; degrade to "no hotplug" with a stderr warning.
- layout.svelte: store the global error/unhandledrejection handler
refs and remove them in onDestroy so HMR/window teardown doesn't
leak listeners.
Verified: cargo check -p kon-core -p kon-storage -p kon-cloud-providers
passes. cargo check on src-tauri/kon-audio/kon-hotkey requires alsa +
gtk system libs not present in this sandbox; their changes are
syntactically and type-checked against the rest of the workspace.
svelte-check requires npm install which is not available here.
https://claude.ai/code/session_018ozAs4UcRC8jbJbddqJtEw
CROSS-PLATFORM AUDIT:
- Linux x86_64 (Fedora 43, KDE Wayland): HIGH confidence (the dev target)
- Linux other distros / X11: MEDIUM (tested patterns, untested distros)
- Windows 10/11: LOW — theoretically supported via Tauri + CPAL + whisper.cpp
+ tauri-plugin-global-shortcut (the custom evdev hotkey is no-op on
Windows); has had zero hands-on testing
- macOS Apple Silicon / Intel: LOW — same; macOS Info.plist needs
NSMicrophoneUsageDescription for the bundled app, and the path bug
fixed in this commit
REAL BUG FIXED — macOS app_data_dir was Unix-style:
crates/storage/src/file_storage.rs::app_data_dir() previously fell into
the "Unix" branch on macOS and wrote to ~/.kon/, which violates Apple
guidelines and confuses install/uninstall tooling. Now correctly:
- Windows: %LOCALAPPDATA%/kon (unchanged)
- macOS: ~/Library/Application Support/Kon/
- Linux: $XDG_DATA_HOME/kon or ~/.local/share/kon (XDG Base Directory),
with legacy ~/.kon/ fallback so existing installs keep working
- Other Unix: ~/.kon/ (unchanged)
OS DETECTION LAYER:
- New get_os_info Tauri command in commands/diagnostics.rs returns:
{os, arch, family, usesCmd, isWayland, customHotkeyBackend,
primaryModifierLabel}
- New src/lib/utils/osInfo.js helper:
- async loadOsInfo() warms a cache, called once at root layout mount
- then isMac() / isWindows() / isLinux() / modKeyLabel() / isWayland()
are synchronous
- browser-preview fallback reads navigator.platform
- Lets components localise hotkey labels (Cmd vs Ctrl), file-picker
copy ("Open Finder" vs "Open Explorer"), etc, without recomputing in
every place.
cargo check -p kon-storage clean. Updated HANDOVER-2026-04-17.md with
a per-platform confidence table.
THE BRIEF:
For the friends beta we need verbal feedback AND technical feedback —
some bugs the user cannot describe but a stack trace can. Built it in
two layers, with Layer 3 deferred until there is real volume.
PRIVACY POSTURE (matches Kon's local-first positioning):
- All capture is to disk only. Nothing is transmitted.
- The manual report bundler shows the user exactly what would be
shared and lets them choose to copy / save it.
- No remote endpoint, no Sentry, no opt-out telemetry.
LAYER 1 — Always-on local capture:
- Rust panic hook (commands/diagnostics.rs::install_panic_hook) writes
each panic to ~/.kon/crashes/<unix-ts>-<short-id>.crash. Captures
thread name, OS/arch, RUST_BACKTRACE state, and the panic info.
Installed before tauri::Builder so it catches setup-time panics too.
- Frontend window.onerror + window.unhandledrejection in
src/routes/+layout.svelte forward to the new log_frontend_error
Tauri command, which inserts into the existing error_log SQLite
table. Best-effort: errors in the error handler itself are swallowed
so logging can never crash the app.
- Existing error_log table (migration v1) is now actually used —
closes the TODO from crates/storage/src/database.rs:409.
LAYER 2 — Manual diagnostic report:
- Settings → About → Diagnostics section: Generate report → Preview →
Copy / Save.
- Report is plain markdown so it pastes cleanly into email, Discord,
GitHub issues. Sections: app version + OS, sanitised settings JSON,
recent error_log rows, last 5 crash dumps with previews, log file
tail (8KB).
- Preview is shown in a <details> with the full text the user can
inspect before deciding to share.
- Save writes to ~/.kon/diagnostic-reports/kon-diagnostic-<ts>.md.
NEW FILES:
- src-tauri/src/commands/diagnostics.rs (panic hook + 5 Tauri commands)
CHANGES:
- crates/storage/src/file_storage.rs: crashes_dir() + logs_dir() helpers
- crates/storage/src/database.rs: ErrorLogRow + list_recent_errors()
- crates/storage/src/lib.rs: re-exports
- src-tauri/src/commands/mod.rs: + diagnostics module
- src-tauri/src/lib.rs: install_panic_hook() before Builder; 5 new
Tauri commands registered (log_frontend_error,
list_recent_errors_command, list_crash_files,
generate_diagnostic_report, save_diagnostic_report)
- src/routes/+layout.svelte: installGlobalErrorCapture() at mount
- src/lib/pages/SettingsPage.svelte: diagnostic state + buttons +
preview block at the end of the About section
cargo check -p kon-storage clean. Settings.svelte if/each balanced.
LAYER 3 (deferred):
- Optional opt-in remote reporting (self-hosted Sentry on Tartarus or
similar). Not for friends beta. Open up after volume justifies it.
Finally exposes the canonical SQLite store to the frontend. Previously
the transcript/task tables existed but no Tauri command read or wrote
them — the UI lived entirely in localStorage. Closes that gap and adds
the missing commands flagged by both Codex and architecture-review.md
§13.
MIGRATIONS:
- v2 adds:
- transcripts_fts virtual table (FTS5, porter+unicode61 tokeniser,
diacritics-folded) backed by transcripts via content_rowid
- INSERT/UPDATE/DELETE triggers keep FTS in sync automatically
- dictionary table for custom vocabulary (term, note, created_at)
- Append-only as required; never modifies v1
STORAGE FUNCTIONS (crates/storage/src/database.rs):
- list_transcripts_paged(limit, offset) — pagination
- count_transcripts() — for "showing X of N" in UI
- update_transcript(id, text?, title?) — closes the historic
architecture-review.md §13 rename-never-persists TODO
- search_transcripts(query, limit) — FTS5 wrapper
- list_dictionary / add_dictionary_entry / delete_dictionary_entry
- DictionaryEntry struct exported
TAURI COMMANDS (src-tauri/src/commands/transcripts.rs new file):
- add_transcript, list_transcripts, count_transcripts_command,
get_transcript, update_transcript, delete_transcript,
search_transcripts
- list_dictionary_command, add_dictionary_entry_command,
delete_dictionary_entry_command
- TranscriptDto + DictionaryDto for camelCase frontend serialisation
REGISTRATION (src-tauri/src/lib.rs):
- All 10 new commands wired into the invoke_handler
cargo check -p kon-storage passes clean. The Tauri crate cargo check
still requires `sudo dnf install cmake clang-devel` for whisper-rs-sys
(pre-existing infra dep, unrelated to this work).
NEXT (still in this sprint):
- Wire DictationPage / FilesPage to call add_transcript on every save
(dual-write: keep localStorage for now, add SQLite alongside)
- Wire HistoryPage rename to call update_transcript (fixes the TODO)
- Add a History search input that calls search_transcripts
- Add Settings → Dictionary panel
- Inject dictionary terms into the LLM cleanup prompt
Closes the 6 Codex review findings on the Day 1 mic-capture work
(commits 96980c7 + 41db162). Detail in
output/reports/kon-codex-mic-capture-followup-2026-04-17.md (CORBEL
workspace).
src-tauri/src/commands/audio.rs:
- D1: Sending an explicit stop signal before dropping stop_tx, so the
accumulator task wakes up immediately rather than waiting for the
Disconnected detection added below.
- D2: Wrapping MicrophoneCapture::start() in tokio::task::spawn_blocking.
start() can spend up to 350ms × N_devices × 2 passes; running it on
the async runtime froze other Tauri commands.
- M3: Match on rx.try_recv() error variants. Empty -> sleep + continue.
Disconnected -> exit accumulator task immediately. Previous behaviour
was an infinite loop after the capture stream died.
crates/audio/src/capture.rs:
- D3: Added DEAD_SILENCE_FLOOR (1e-7) gate that applies even in the
fallback pass. PulseAudio/PipeWire can stream zero-valued bytes from
a borked device; that is worse than failing fast.
- M1: Validation requeue (`for chunk in collected { try_send }`) now
counts drops in the same dropped_chunks counter.
- M2: New CaptureRuntimeError type. The cpal stream error callback now
forwards errors via a separate sync_channel (capacity 16) that the
live session can drain via take_error_rx() and surface as toasts.
Re-exported from kon_audio crate root.
cargo check -p kon-audio passes clean.
NOT YET DONE (M4): JACK-specific monitor name patterns. Defers until
testing on a JACK host. Current is_monitor_name() may miss JACK
conventions.
Wiring the new error_rx into the live session and surfacing as toasts
lands with the Day 3 error-toast system commit.
Day 1 of the upgrade plan (output/reports/kon-upgrade-plan-2026-04-17.md
in the CORBEL workspace). Fixes the HANDOVER.md blocker: native live
transcription was capturing silence because PulseAudio/PipeWire monitor
sources (speaker loopback) were winning the device-selection race —
they deliver zero-valued bytes that satisfied the original
"any device that produces data within 350ms" check.
WHAT CHANGED:
crates/audio/src/capture.rs (rewrite):
- New `DeviceInfo` struct (serde-derived) for the Settings device picker
- New `MicrophoneCapture::list_devices()` enumerates inputs with metadata
- New `MicrophoneCapture::start_with_device(name)` for explicit selection
- Refactored `start()` with monitor-source filtering by name pattern
(.monitor suffix, "Monitor of " prefix, "loopback" substring) and
RMS-energy validation in a 350ms window
- Two-pass selection: real inputs first, monitor sources only as
last-resort fallback with explicit warning log
- Drop counter (Arc<AtomicU64>) tracks chunks dropped by `try_send`
failure under load — Codex review caught this as a silent-failure risk
- `dropped_chunks()` accessor for the live session
- Verbose tracing at every step for diagnosis
- Unit test for monitor-name detection
- `cargo check -p kon-audio` passes clean
crates/audio/src/lib.rs:
- Re-export `DeviceInfo`
crates/audio/Cargo.toml:
- Add serde dependency (for DeviceInfo derives)
src-tauri/src/commands/audio.rs:
- New `list_audio_devices` Tauri command (returns Vec<DeviceInfo>)
src-tauri/src/lib.rs:
- Register `list_audio_devices` in invoke_handler
src/lib/pages/SettingsPage.svelte:
- New "Audio" section with microphone picker dropdown
- Auto-populates on mount via `list_audio_devices`
- Refresh button + clear messaging about monitor sources
- Likely-monitor entries marked disabled in the dropdown
- Auto mode is the default (empty string in settings.microphoneDevice)
src/lib/stores/page.svelte.js:
- New `microphoneDevice` field in defaults (empty = auto-select)
NEXT STEPS (per the upgrade plan):
- Wire `microphoneDevice` from settings into `MicrophoneCapture::start_with_device`
in the live and standalone capture paths (currently both still call
the auto-selecting `start()`)
- Test on real hardware (Wayland + multiple input devices)
- Codex sanity-check of this diff is running in parallel; addendum to
follow if anything substantive comes back
Refs: /home/jake/Documents/CORBEL-Projects/kon/HANDOVER.md
output/reports/kon-upgrade-plan-2026-04-17.md (CORBEL workspace)
The CTC model (onnx-community/parakeet-ctc-0.6b-ONNX) only has an encoder
— no decoder_joint or nemo128 preprocessor. transcribe-rs expects a TDT
transducer variant with all three. Switch to istupakov/parakeet-tdt-0.6b-v2-onnx
which has the correct int8 files.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Upstream transcribe-rs 0.3.10 added required fields to TranscribeOptions.
Set both to None (use engine defaults).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add kon-hotkey crate with evdev-based global hotkey capture that works on
Wayland (and X11). Patterns from whisper-overlay: per-device async listeners,
inotify hotplug with udev permission retry, watch channel for live config
updates. Frontend detects Wayland at startup and selects evdev or
tauri-plugin-global-shortcut automatically.
Model downloads now support HTTP Range resume for interrupted downloads and
optional SHA256 integrity verification (incremental, no second pass).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
