Storage-side scaffolding for B3.3 (granularity prompt variants) and
B3.10 (mastery fade). The user-facing wiring (decompose_and_store
granularity threading, complete_subtask_cmd count bump, MicroSteps
inline prompt + skip-check, Settings UI) is part 2 of this batch.
B3.3 (prompts.rs): three new system-prompt constants
DECOMPOSE_LIGHT_SYSTEM (3 atomic verb-first steps),
DECOMPOSE_DEFAULT_SYSTEM (4-5 balanced steps), DECOMPOSE_DETAILED_SYSTEM
(6-7 with brief context) all preserve the cue-anchored "When [cue],
[action]" framing from PR 1.1. DECOMPOSE_TASK_SYSTEM remains as an
alias of Default for back-compat with existing callers and tests.
4 snapshot tests assert the framing survives across variants and the
alias matches.
B3.10 (storage): migration v18 adds microstep_patterns
(normalized_title PK, sample_title, completed_count, skip_breakdown,
prompted_at) plus an index on completed_count. New storage functions:
normalize_microstep_title (lowercase + whitespace-collapse + trim),
get_microstep_pattern, upsert_microstep_pattern_increment,
set_microstep_pattern_decision. 5 storage tests cover normalisation,
upsert/bump semantics, decision persistence, and the threshold query.
74 storage tests pass (was 69), 9 prompts tests pass (was 5).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Settings: 8 new fields (lastLaunchAt, reentryFreshStartUntil,
lastActiveProfileId, microStepGranularity, nudgeMode, nudgeDigestTimes,
sparklineRangeDays, energyLabels) with sane defaults; nudgesEnabled
auto-maps to nudgeMode on first load (true→immediate, false→off).
Profile: optional energyLabelsOverride field on the localStorage
Profile shape — when present, overrides the global energyLabels;
absent/null falls through to settings.energyLabels via the new
resolveEnergyLabels helper.
Templates: migration v17 adds the templates table (id PK, name,
sections JSON, created_at, updated_at). 5 storage CRUD functions
plus import_templates with idempotent duplicate-id handling. 5 Tauri
commands (list/create/update/delete/import) wired and registered.
Frontend store rewired to read from SQLite via list_templates_cmd;
one-time migration imports kon_templates localStorage and clears it
on success. Fresh installs only seed the new {Meeting notes, Daily
check-in} defaults. Existing user templates survive the migration
verbatim.
Test coverage: 5 new Rust tests (CRUD, import idempotency, atomic
failure, updated_at bump, missing-id delete is no-op).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Storage: 5 task_lists CRUD functions (list/create/update/delete/
import) and 4 archive functions (list_archived_tasks/archive_task/
unarchive_task/archive_inbox_older_than). 8 new tests cover CRUD,
import idempotency, atomic-failure rollback, dependent-task null-out,
archive scope, archive_inbox_older_than Inbox-only invariant, and
unarchive round-trip.
Tauri: 9 new commands (5 task_lists + 4 archive) registered in the
invoke_handler. New file commands/task_lists.rs mirroring tasks.rs
structure. TaskDto extended with archived/archivedAt.
Frontend: loadTaskLists rewired to read from SQLite via
list_task_lists_cmd on Tauri runtime, with a one-time migration
that imports kon_task_lists localStorage into SQLite via
import_task_lists_cmd and clears localStorage only on success.
mapTaskRow handles the new archived fields. TasksPage gets an
Archived filter pill (hidden when empty) and a per-row archive
icon. moveTaskList/sortTaskLists kept local pending a future
order-column migration.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Migration v16 adds tasks.archived (default 0) + tasks.archived_at +
idx_tasks_archived. TaskRow extended with both fields; list_tasks and
list_subtasks default-filter archived rows; complete_subtask_and_check_parent
SELECT lists updated to round-trip the new columns. TaskListRow +
ImportSummary structs land in preparation for the task_lists CRUD
that part 2 wires up. All 61 existing tests still pass.
(SQL splitter is naive about ';' inside -- comments, so the v16
comment block was rewritten to avoid a bare semicolon mid-comment
that the splitter was treating as a statement boundary.)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Change complete_subtask_and_check_parent to return
(TaskRow, Option<TaskRow>): the refreshed subtask, plus the parent when
the cascade auto-completed it. Both rows are read inside the same
transaction so the frontend never observes partial state.
Wrap the cmd in a CompleteSubtaskResult { updatedSubtask,
autoCompletedParent } DTO. MicroSteps now applies the refreshed subtask
locally and, when present, calls a new replaceTaskFromDto helper on the
Tasks store so the parent's done/doneAt/badge counts update without a
follow-up list_tasks_cmd round-trip.
Test: complete_subtask_returns_updated_subtask_and_optional_parent
covers both paths (sibling-pending → None, last-sibling → Some(parent)).
All 61 kon-storage tests pass; svelte-check clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add cue-anchored "When [cue], [action]" framing to the task-decomposition
prompt where natural cues are present (Gollwitzer-style implementation
intentions, d=0.65 effect size). Soften Bionic Reading and accessibility-
font copy to honest preference framing per the v3 audit (Strukelj 2024;
Doyon n=2,074). Update timer nudge from "Still on that timer?" (which
read as judgmental) to "Timer's still running." Replace stale Tasks
page header copy promising automatic extraction.
Audio envelopes (focusTimer 20ms ramp, sounds.ts 10ms attack) verified
correct per memo §B; no code change needed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Recorder-style auto-titling for transcripts. Mirrors the Phase 9
content-tags pipeline so the same prompt-injection-hardened pattern,
spawn_blocking discipline, and sanitisation-after-generation shape get
reused; the user-facing surface (auto on save + on-demand button) lands
in a follow-up commit.
- crates/llm/src/prompts.rs: new TRANSCRIPT_TITLE_SYSTEM constant. Same
injection guard wording as ai-formatting's CLEANUP_PROMPT — dictated
speech is data, not instructions. Rules constrain output shape: 4-8
words, Title Case, no quotes, no terminal punctuation, "Untitled"
fallback for empty input.
- crates/llm/src/lib.rs: LlmEngine::generate_title returns
Result<String, EngineError>. Mirrors extract_content_tags shape:
trailing-2000-char UTF-8-boundary truncation, temperature 0,
max_tokens 24, free-form output (no GBNF — titles are prose, not a
closed set). Sanitisation runs server-side via the new private
sanitize_title helper, which handles the real Qwen3 failure modes:
surrounding curly + ASCII quotes, leading "Title:" prefix, multi-line
output, trailing "." / "!" / "?", whitespace runs, 100-char cap,
literal "Untitled" → None. Three unit tests cover composite real-world
outputs end-to-end. kon-llm test suite goes 15 → 18 passing.
The Tauri wrapper, invoke_handler registration, and frontend wiring
follow in subsequent commits.
Two issues in flush_is_idempotent_and_leaves_clean_state from
581a098:
1. silence_close_samples and max_chunk_samples were cast `as u64`
but with_thresholds takes usize — wouldn't compile.
2. enter_threshold was 0.005 and exit_threshold 0.01, which
violates the hysteresis invariant (enter must be >= exit) and
panics in debug_assert at runtime. Swap to 0.01 / 0.005 so the
test actually runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both `kon-transcription` and `kon-llm` previously hardcoded their native
acceleration features in Cargo.toml — `whisper-rs` with `vulkan`,
`llama-cpp-2` with `openmp` + `vulkan`. That worked everywhere desktop
ships (Linux/macOS/Windows all have Vulkan via MoltenVK on Mac), but it
made an Android build structurally impossible: NDK builds against drivers
that vary wildly across SoCs (Adreno OK, Mali patchy, PowerVR worse), and
some older devices have no Vulkan at all.
Roadmap step 0 from the Android plan: make the GPU acceleration
opt-in so a CPU-only target compiles. Reuses the existing pattern that
README's "future Windows non-AVX2 build" comment hinted at.
- kon-transcription: new `whisper-vulkan` feature gates `whisper-rs/vulkan`
via the optional-syntax `whisper-rs?/vulkan`. Default features stay as
`["whisper", "whisper-vulkan"]` so desktop is unchanged.
- kon-llm: new `gpu-vulkan` and `openmp` features each gate the matching
`llama-cpp-2` feature. Default stays `["gpu-vulkan", "openmp"]`. They are
independent so an Android Vulkan build can opt into vulkan without
openmp (NDK OpenMP linking has known cross-version fragility).
CPU-only build invocations:
cargo build -p kon-transcription --no-default-features --features whisper
cargo build -p kon-llm --no-default-features
Verified: all 91 tests in the buildable-in-sandbox crates still pass.
The two crates whose Cargo.toml changed (kon-transcription, kon-llm)
can't be compiled in this sandbox (ort-sys CDN + cmake-built llama.cpp);
CI's Linux/macOS/Windows builders will exercise the default-feature path
exactly as before.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
The evdev listener's run loop did `let _ = event_tx.send(event).await`
inside the trigger-key match arm. If the receiver was dropped without
the explicit shutdown signal (set hotkey to None), the send returned
Err and the loop kept polling — sending into a closed channel forever
until something else terminated the task.
Replace with explicit handling: on Err, log via log::warn! once and
return Ok(()) from `run`. The shutdown-via-None path is unaffected.
kon-hotkey still 4/4.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
The cpal stream-error closure used `let _ = err_tx.try_send(...)` against a
bounded sync_channel(16). If the live session's listener stalled or the
frontend disconnected, runtime stream errors were silently dropped — the
diagnostic bundle showed nothing for a session that mysteriously stopped
working.
- Bump the error channel capacity 16 → 32 (matches AUDIO_CHANNEL_CAPACITY).
- On try_send failure, log to stderr with the device name + a per-session
drop counter so the symptom is visible in the diagnostic bundle even
when the typed event never reached the frontend.
- Plumb a new `dropped_errors: Arc<AtomicU64>` through `build_input_stream`
alongside the existing `dropped_chunks`, mirroring the same pattern.
(kon-audio doesn't build in the audit sandbox: it links against ALSA
which the sandbox lacks. CI cross-platform compiles it.)
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
The earlier audit noted that `flush()` had three exit paths but only two
of them explicitly cleared all state-machine fields:
1. InSpeech with non-empty active_chunk: emit_active_chunk_and_close()
handled it.
2. InSpeech with empty active_chunk (hit_max-mid-flush): handled inline.
3. Idle (no padded frame, or padded frame closed cleanly): no explicit
reset — silent_tail_samples / pending_onset_frames / onset_buffer
could carry stale values from `consume_frame` calls inside the same
flush.
In the worst case, the first feed of a fresh recording could see leftover
onset bookkeeping and produce a chunk start that doesn't match the new
session's audio. Reusing the same `RmsVadChunker` across stop/start is
the main path that would hit this.
Add a single defence-in-depth reset block at the end of flush — every
exit path lands the chunker in the same fields a fresh chunker has,
except `next_sample_index` (the running total-samples counter, intent-
ionally preserved). Test asserts: a second flush after a full speech →
silence → partial-pending sequence emits zero chunks, and a subsequent
silent feed also emits zero, proving no stale state leaked.
(kon-transcription doesn't build in the audit sandbox because ort-sys's
build script can't reach pyke's CDN; CI cross-platform compiles it.)
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
The error_log table had no retention policy: every backend error was
appended forever, so across months of dogfooding it grew unbounded. That
silently bloats the diagnostic-bundle export and slows the
list_recent_errors query the Settings → About panel runs.
- New `kon_storage::prune_error_log(pool, keep_days)` does a single
`DELETE FROM error_log WHERE timestamp < datetime('now', '-Nd days')`
and returns the row count removed.
- src-tauri/src/lib.rs runs it once during setup() with a const
ERROR_LOG_RETENTION_DAYS = 90. Failure is logged to stderr but does not
block startup — a prune that fails is strictly less important than the
app coming up.
- Test: insert three rows at now / -30d / -200d, verify a 90-day prune
removes only the oldest, and a subsequent 14-day prune removes the
-30d row. Storage suite at 60/60.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
`detect_meeting_processes` is called every 15 s when meeting-auto-capture
is enabled. The previous `list_running_process_names` allocated a fresh
`sysinfo::System` per call and walked /proc cold; on a busy host
(~300 processes) that's ~50–100 ms of work, every poll, forever.
Add `kon_core::process_watch::ProcessLister`, a thin wrapper around a
long-lived `System` whose process table is refreshed in place. The Tauri
host holds one behind a `Mutex<ProcessLister>` in a new `MeetingState`
managed at app setup. The free `list_running_process_names` is kept as a
convenience that constructs a fresh `ProcessLister` per call — its only
remaining caller is the existing smoke test.
- ProcessLister + Default in crates/core/src/process_watch.rs.
- MeetingState in src-tauri/src/commands/meeting.rs; the command takes
it via `tauri::State` and locks for the duration of the snapshot.
- src-tauri/src/lib.rs registers MeetingState alongside the other
managed states.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
Migration v15 adds a composite index covering the dominant transcripts
query path:
SELECT ... FROM transcripts
WHERE profile_id = ? ORDER BY created_at DESC LIMIT ?
Previously SQLite had to choose between idx_transcripts_profile_id
(filter by profile, then in-memory sort by date) and idx_transcripts_created
(scan dates and filter on profile). Both work fine at hundreds of rows
and degrade past a few thousand.
`migration_v15_creates_profile_created_index` asserts (a) the index exists
and (b) `EXPLAIN QUERY PLAN` shows the planner picks it for the canonical
profile-scoped, date-ordered list query.
Test count assertions in `test_migrations_run_on_empty_db` and
`test_migrations_idempotent` bumped 14 → 15.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
The kon-mcp stdio server is documented as "read-only, no auth, local-only"
but until now opened the SQLite store via `kon_storage::init`, which returns
a writable pool and runs migrations. Read-only-ness was enforced only by the
exposed tool surface (list_transcripts, get_transcript, search_transcripts,
list_tasks); a future bug or a malformed dispatch could escape into a write
against the user's primary database.
Add `kon_storage::init_readonly` that opens with `SqliteConnectOptions
::read_only(true)` and `create_if_missing(false)`, no migrations. The
constraint is now structural — SQLite rejects writes at the connection
level regardless of which handler runs.
- New `init_readonly(path)` in crates/storage/src/database.rs.
- Re-exported from kon_storage.
- crates/mcp/src/main.rs switched over and updated startup banner.
- Two tests: writes fail on the read-only pool, reads succeed; opening a
non-existent DB returns an error instead of silently creating one.
https://claude.ai/code/session_0189xUb6ie6t9qHkzatGZ9Rb
Adds llm_tags TEXT NOT NULL DEFAULT '' to the transcripts table via
new migration v14. SELECT statements + TranscriptRow + transcript_row_from
now carry the column. update_transcript_meta gains a sixth Option for
llm_tags following the existing COALESCE pattern; an
#[allow(too_many_arguments)] keeps clippy happy without inverting the
signature into a struct that would just shift the indirection.
The Tauri-side TranscriptDto + UpdateTranscriptMetaRequest + the
update_transcript_meta_cmd command pass llm_tags through unchanged.
Pre-existing manualTags persistence path now has a sibling for
llmTags ready for the frontend to call.
Phase 8 brittle test fix included: list_recent_completions_uses_local_day_boundary
was anchoring its "-2 days" UTC offset against the local-day spine,
which drifted across UTC midnight. Anchored to the local date directly
so it matches the spine regardless of clock.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Added as a method on LlmEngine alongside cleanup_text and
extract_tasks; same render_chat_prompt -> generate -> parse pattern.
Truncates the transcript to its trailing 2000 chars on a UTF-8 char
boundary, runs at temperature 0.0 with the CONTENT_TAGS_GRAMMAR GBNF,
and re-validates intent against INTENT_CLOSED_SET to catch the
unlikely grammar bypass case. max_tokens 96 is enough for the JSON
envelope. Smoke test gated on KON_LLM_TEST_MODEL like the existing
smoke.rs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ContentTags serde-serialisable. CONTENT_TAGS_SYSTEM is the system
message rendered at extraction time; INTENT_CLOSED_SET is the single
source of truth for the enum values the grammar restricts. Grammar is
strict: lowercase hyphen-joined topic 3+ chars (max enforced by
max_tokens at call site), intent from the closed set, JSON-only
output. Recursive topic-rest matches the existing GBNF style in this
file.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Returns a fixed-length, oldest-first series of daily completion counts
for the last N local-time days. Excludes cascade parents and
uncompleted rows. Empty days are explicit zeros, not missing entries,
so the Phase 8 sparkline can render a fixed 7 bars.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
uncomplete_task now clears auto_completed alongside done / done_at on
both the target row and the cascaded-parent reopen. Keeps the flag
accurate so a later re-completion via a different path is counted
correctly by the Phase 8 daily-count query.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Matches the project's style preference for full stops over em/en dashes.
No functional change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
complete_subtask_and_check_parent now sets auto_completed = 1 on the
parent when it closes via the cascade. The subtask UPDATE itself
remains at the default 0, so explicit user taps still count toward
the Phase 8 daily total.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a flag on tasks to distinguish manual completions from the
cascade auto-completion performed by complete_subtask_and_check_parent.
Partial index on (done_at, auto_completed) supports the Phase 8
daily-count query without bloating the tasks index footprint.
Forward-only: pre-migration completed rows default to 0 (they count).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Small if-then automation layer. Rules persist in SQLite; the runner
lives on the frontend and binds to the Phase 6 event bus so the rule
pipeline reuses the same delivery primitives (timer events, TTS,
Tasks navigation).
Storage:
- Migration v12 adds implementation_rules (id, enabled, trigger_kind,
trigger_value, actions_json, last_fired_key, created_at,
updated_at) with enabled+trigger_kind index for the runner's hot
path.
- CRUD helpers: insert / list / get / set-enabled / mark-fired /
delete, plus a round-trip test.
Commands (all main-window-guarded via ensure_main_window):
- list_implementation_rules
- create_implementation_rule — validates HH:MM, checks the target
task exists at save time for surface-task actions, caps
speak-line at 240 chars, pins v1 timers to 5 minutes.
- set_implementation_rule_enabled
- mark_implementation_rule_fired — main-thread idempotency shim so
the runner can atomically claim a fire.
- delete_implementation_rule
Runner (implementationIntentions.svelte.ts):
- Subscribes to kon:task-completed and kon:morning-triage-finished
(MorningTriageModal now emits on all three exit paths — empty,
skipped, picked — so skip counts as finishing).
- 30 s poll for time-of-day rules, plus an immediate check on startup
so a rule whose time has already passed today catches up once.
- Idempotency via last_fired_key composed as YYYY-MM-DD@HH:MM for
time rules; new time rules whose HH:MM has already passed today
are pre-seeded so they don't fire retroactively on save.
- Rules are paused when Nudges "Mute for now" is on — a hard mute
stops all rule delivery in addition to OS notifications.
- Stale-task safety: if a surface-task action's target has been
deleted, the runner opens Tasks and warns clearly rather than
pretending to surface something that's gone.
Editor (ImplementationRulesEditor.svelte):
- Lives in Settings under a new "If-then rules" accordion section.
- `If` picker: time of day (with time input), a task completes,
morning triage finishes.
- `Then` composer: optional surface (inbox / today / all tasks /
specific task), optional 5-min timer, optional speak-aloud line.
- Saved rules list with enable toggle + delete.
Rules table integration for Phase 10b rename sweep: add
implementation_rules to the kon.db → corbie.db migration shim when
that phase lands.
Gates: fmt, clippy -D warnings, cargo test 265/0, svelte-check
0/0, npm build green. Pre-existing Vite chunk warning on sounds.ts
is unrelated to Phase 7.
Closes Phase 3 of the 2026-04-23 feature-complete roadmap. Incorporates
the Codex plan-review fixes from this session: profile-free index, tri-
state update command, and de-prioritise-not-hide semantics.
Storage (kon-storage):
- Migration v11 adds `energy TEXT` to `tasks` with a CHECK constraint on
`high | medium | brain_dead | NULL`. Index `(energy, created_at DESC)`
— deliberately not per-profile because the tasks table carries no
profile_id column yet (tracked as a separate gap in HANDOVER).
- `TaskRow.energy: Option<String>` plus `task_row_from` read.
- `insert_task` signature grows by one optional arg (`energy`). Allowed
`too_many_arguments` with a rationale comment — the positional shape
matches the column order and flipping to a params struct would have
rippled through every caller for cosmetic benefit only.
- New `set_task_energy(pool, id, Option<&str>) -> TaskRow`. Lives as its
own function because `update_task` uses COALESCE to let `None` mean
"preserve" — which would make clearing the tag impossible.
- Two new tests: round-trip including explicit NULL clear, and CHECK
constraint rejection of unknown values.
- Tests updated for the v10 → v11 version bump.
Tauri (src-tauri):
- `TaskDto.energy`. `CreateTaskRequest.energy` (optional). Inline
validation against the allowed set before hitting the DB, so frontend
bugs surface as friendly errors instead of CHECK-constraint failures.
- New `set_task_energy_cmd` command mirroring the storage tri-state API.
Frontend (svelte):
- `EnergyLevel` type added to `types/app.ts`. `TaskDto`, `TaskEntry`, and
`TaskDraft` grow an `energy` field.
- `SettingsState.currentEnergy` (persisted) + `matchMyEnergy` (persisted
toggle). Defaults: null + false — no surface change until user opts in.
- `setTaskEnergy(id, EnergyLevel | null)` action on the task store.
Calls the dedicated Tauri command, updates local state, broadcasts to
sibling windows.
- `EnergyChip.svelte` — new component. Cycles unset → High → Medium →
Brain-Dead → unset on click. Colour tokens: accent / warning /
text-tertiary (deliberately not danger-red for Brain-Dead — the brief
is explicit that this state must not feel pathologised).
- Chip rendered on every task row in TasksPage and every row in
WipTaskList. Hidden-until-hover when energy is unset so untagged rows
stay calm; always visible once tagged because the colour is the signal.
- Tasks page header gains a "I feel" segmented control and a
"Match my energy" toggle. When both are active, matching tasks sort
to the top — unset tasks are treated as Medium-equivalent. Nothing is
ever hidden; this is a de-prioritisation, not a filter.
Deferred / out of scope:
- LLM-driven surfacing (brief says "The AI surfaces...") — deterministic
client-side sort is v1; LLM layer is a later phase.
- tasks.profile_id + per-profile energy sort — separate migration.
All green: cargo build + 251 tests + clippy -D warnings (0 warnings)
+ fmt + svelte-check (0/0) + npm run build.
Independent review surfaced three majors and one nit. All actioned.
MAJOR 1 — profile scoping:
`decompose_and_store` and `extract_tasks_from_transcript_cmd` now
accept an optional `profile_id` (wired from `profilesStore.activeProfileId`
in MicroSteps.svelte and DictationPage.svelte), and thread it into the
feedback-retrieval query so per-profile decomposition styles do not
leak into each other. `record_feedback` gets the same treatment.
MAJOR 2 — prompt-budget regression on long inputs:
New `trim_to_budget` helper + `FEW_SHOT_CHAR_BUDGET = 2000` char cap
in `src-tauri/src/commands/tasks.rs`. Retrieval still pulls up to 5
rows but they are char-counted and truncated against the budget
before being sent to the LLM. Char cost matches the `Input: ...\n
Good output: ...` render path so the budget maps cleanly to ~570
Qwen3 tokens, well inside the 8192-context reserve after the 512-
or 768-token response allocation. Oldest-first drop order (iteration
stops at cost exceeded) preserves the most recent correction which
is the one carrying the user's live preference.
MAJOR 3 — inline edit stale-rollback race:
`saveEdit` in MicroSteps.svelte now stamps a monotonic per-step
`saveToken`. Each edit bumps the token; on failure the rollback
only fires if `saveToken[step.id] === myToken`, so a slow-failing
first save can no longer overwrite a faster successful second save.
NIT — retrieval ordering stability:
`list_feedback_examples` ORDER BY now `created_at DESC, id DESC`.
SQLite timestamp precision is one second; without the secondary
key, bursty feedback within the same second would select
non-deterministically.
Also: malformed `context_json` now warns via eprintln! rather than
disappearing silently — Codex minor.
All green: cargo build + 249 tests + clippy -D warnings + fmt
+ svelte-check (0/0) + npm run build.
Closes the human-in-the-loop gap from docs/brief/feature-set.md and
Phase 2 of the 2026-04-23 feature-complete roadmap.
Storage (kon-storage):
- Migration v10 adds the `feedback` table: (target_type, target_id,
rating, original_text, corrected_text, context_json, profile_id,
created_at) with CHECK constraints on target_type and rating, plus
indexes on (target_type, rating, created_at DESC) for prompt-time
retrieval and (profile_id, target_type, created_at DESC) for
per-profile scoping.
- New public API: `FeedbackTargetType`, `RecordFeedbackParams`,
`FeedbackRow`, `record_feedback`, `list_feedback_examples`.
- Tests updated — the RB-02 rollback regression now discovers the
real max version at runtime instead of hard-coding v10 for its
poison migration.
LLM (kon-llm):
- `prompts::FeedbackExample` — local shape for few-shot exemplars so
kon-llm stays independent of kon-storage.
- `prompts::build_conditioned_system_prompt` — appends a "here is
the style this user prefers" block to the base system prompt
when examples are available; returns the base prompt unchanged
when empty, so new users and early sessions see generic output.
- `LlmEngine::decompose_task_with_feedback` and
`LlmEngine::extract_tasks_with_feedback` thread examples through
to the builder. The old one-arg variants are preserved and now
call through with an empty slice.
- 4 unit tests covering empty, empty-input-skip, correction-wins,
and thumbs-up-only fallback.
Tauri (src-tauri):
- New commands::feedback module: `record_feedback`,
`list_feedback_examples_cmd`.
- `decompose_and_store` and `extract_tasks_from_transcript_cmd`
now fetch the last 5 positive/neutral feedback rows for their
target type and pass them through to the LLM, wiring the
learning loop end-to-end.
- Shared `to_llm_examples` helper parses the `context_json.input`
field (where the recorder stashes the parent task text / transcript
chunk) back into the exemplar shape.
Frontend (MicroSteps.svelte):
- Thumbs-up and thumbs-down buttons on every micro-step row.
Hover-revealed; the vote recolours the icon; clicking again
clears the local highlight (the row itself stays in the audit
trail).
- Pencil icon + double-click to edit step text. Save flows through
update_task_cmd for persistence and records a correction feedback
row with (original_text, corrected_text) — the highest-value
training signal.
- Parent task text is captured in context_json.input at record time
so the prompt builder can reconstruct the (input, preferred-output)
pair on subsequent decompositions.
- Feedback capture is best-effort — a record_feedback failure never
interrupts the primary action.
What's deferred to a later phase:
- Thumbs + corrections on extracted tasks (same pipeline, different
surface — probably TasksPage after the AI-extraction path)
- Thumbs on transcript cleanup output
- Semantic retrieval over the feedback corpus (once there is enough
data to justify embedding infrastructure; the storage shape is
already ready for it)
Auto-applied cargo clippy --fix across 11 files — needless return,
unnecessary cast, map_or simplification, repeat().take() → repeat_n(),
iter().any() → contains(), manual char comparison, lifetime elision,
push_str single-char, reference immediately dereferenced.
Also fixed three lints on file_storage.rs manually: two doc-list-item
overindentations, plus the same needless-return. Baseline main was
not clippy-clean with -D warnings before; after this pass one
needless_range_loop warning remains (live.rs:1089) that clippy's
suggested rewrite would make less readable — left for a dedicated
refactor session.
Build + workspace tests remain green (245 passing, 0 failing, 1
ignored).
Every multi-statement migration and its matching schema_version insert
now execute on the same sqlx Transaction. A failure anywhere — a bad
statement, the version insert, or the commit itself — rolls the
database back to its previous state, so the next startup replays the
migration against a clean schema rather than a half-mutated one.
Extracted run_migrations_slice(pool, migrations) as the single apply
path. run_migrations delegates to it with MIGRATIONS; the test helper
run_migrations_up_to now filters MIGRATIONS by target and delegates to
the same code, eliminating the duplicated loop that previously lived
in the test module.
Regression test multi_statement_migration_rolls_back_on_failure
injects a poisoned v9 migration (valid CREATE followed by a bogus
function call) and asserts neither the partial schema change nor the
schema_version row persists after the failure.
SQLite DDL participates in transactions, so this is sufficient. Any
future migration that needs an implicitly-committing statement
(VACUUM / REINDEX / ATTACH — none today) must be its own
non-transactional migration; that's a reviewer responsibility.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
decode_audio_file's packet loop was `Err(_) => break`, so any non-EOF
read error during playback dropped out silently with whatever samples
had accumulated. Per-packet decode errors were tallied and skipped,
contributing to the same outcome. A corrupt or truncated input
therefore came back as `Ok(partial_samples)` — no way for callers to
distinguish a clean decode from a compromised one.
Every SymphoniaError other than the explicit EOF
(`IoError(UnexpectedEof)`) now maps to `AudioDecodeFailed`. Decoder
errors bubble via `?` rather than being counted. `ResetRequired`
promotes to an error rather than a silent break.
Extracted an internal `decode_media_stream(mss, hint)` so tests can
inject a custom `MediaSource`. Added `FlakyCursor` — a seekable cursor
that returns a synthetic I/O error after N bytes — and a regression
test that confirms mid-stream read failure surfaces as `Err` instead of
returning partial audio. Happy-path and missing-file tests added for
coverage.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
try_attach_device was rejecting any device that did not report KEY_A or
KEY_R — a leftover heuristic from the whisper-overlay seed. A user whose
binding was anything else (Ctrl+Shift+D is a common default) would see
no hotkey events from that device even though it supports the key.
Replace the hard-coded check with device_supports_combo(supported,
combo), a pure helper that reads the configured trigger key code from
the HotkeyCombo snapshot. Snapshot is taken from hotkey_rx.borrow()
before opening the device; an unconfigured or shutting-down listener
short-circuits to a non-attach.
Four regression tests in linux::tests cover: supported+D → attach,
unsupported → reject, no reported keys → reject, and the explicit
non-A/non-R case that demonstrates the bug.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Regression surfaced by the batch review: commit 8400128 switched
list_transcripts from unwrap_or_default to map_err(-32602). This
correctly errors on malformed payloads but also rejected the common
case where a client omits the 'arguments' field entirely — which
arrives as Value::Null, and serde_json::from_value does not
deserialise Null into a struct.
Short-circuits the Null case to Args::default() before attempting
deserialisation. Genuine shape mismatches ("limit": "twenty")
still return -32602 as the previous test asserts.
New regression test: tools/call with list_transcripts and no
arguments key must return a successful response.
2026-04-22 review MINORs and NITs:
- crates/core/src/providers.rs: delete entire module. SpeechToText /
TextProcessor / ProviderRegistry were forward-looking traits that
never got wired — the Transcriber trait in kon-transcription
(A.2 #13) has since superseded SpeechToText, and the Registry
pattern was redundant against LocalEngine. Keeping them as dead
public surface signalled future direction that is no longer
accurate.
- crates/core/src/types.rs: delete TranscriptMetadata. Forward-
looking struct with an unfulfilled TODO; storage has evolved
independently through v7/v8 migrations without adopting it.
- crates/ai-formatting/src/llm_client.rs: remove #[allow(dead_code)]
from CLEANUP_PROMPT and format_dictionary_suffix. Both are
actively called; the suppressions would hide future genuine
dead-code warnings in this regression-sensitive prompt file.
- src-tauri/src/commands/live.rs: remove #[allow(dead_code)] from
LiveStatusMessage. Every variant (Warning, Overload, Error,
Finished) is constructed in the module today.
- README.md: update kon-transcription row from "SpeechToText
trait" to "Transcriber trait" and mention the new streaming/
module.
Workspace test gate green (225 lib tests across all crates).
MAJOR from the 2026-04-22 review (crates/mcp/src/lib.rs:188-195):
the handler called serde_json::from_value(args).unwrap_or_default(),
so a request like { "limit": "twenty" } silently became the default
limit of 20. Every other tool handler in this file map_errs to
-32602 Invalid arguments; this one was the outlier.
Switches to the same map_err pattern. Empty params still
deserialise cleanly to Args::default (via #[serde(default)] on the
Option<i64> field), so callers that send no args are unaffected —
only genuinely malformed shapes now error.
Regression test: tools/call with list_transcripts and a
string-typed limit must return code -32602 with an "Invalid
arguments" message.
MAJOR from the 2026-04-22 review (crates/mcp/src/main.rs:26-30): the
stdio transport logged malformed JSON lines to stderr and continued
without sending any JSON-RPC response. Clients saw silence instead of
the -32700 Parse Error they could key off. handle_message has a
parse-error branch for shape mismatch, but it never ran for bytes
that failed to parse as JSON at all.
Exposes a new public helper kon_mcp::parse_error_response(detail)
that mirrors the existing internal error_response pattern, filling
id with null per JSON-RPC 2.0 §5.1 (parse error, no id recoverable).
main.rs now writes that response out before continuing the read
loop.
Regression test on the helper asserts: jsonrpc "2.0", id null,
code -32700, message starts with "Parse error" and includes the
underlying serde detail.
MAJOR from the 2026-04-22 review (database.rs:389-449):
complete_subtask_and_check_parent auto-completes a parent task when
the last child completes, but uncomplete_task only flipped the
requested row — reopening a child left the parent wrongly marked
done, breaking the "parent done iff every child done" invariant.
Wraps uncomplete_task in a transaction and, after flipping the
subtask, looks up its parent_task_id. If present, resets the
parent to done=0 as well. Scoped to "done=1" on the parent update
so an already-open parent is untouched.
Two regression tests:
- uncomplete_subtask_reopens_auto_completed_parent: the direct
mirror of the existing subtask_crud_roundtrip completion flow.
- uncomplete_top_level_task_does_not_touch_siblings: ensures the
parent-reopen branch is a no-op for tasks with no parent, and
siblings without a parent relationship are unaffected.
MAJOR from the 2026-04-22 review (model_manager.rs:161-262): reqwest
does not return Err on 4xx/5xx by default. The resume branch
validated 206/200 and errored on anything else, but the non-resume
branch skipped the status check entirely — a 404 or 500 body was
streamed into .part and atomically renamed over the destination as
if it were the model file. For models without a sha256 declared,
this failure mode was silent and catastrophic (the engine would
crash loading an HTML error page as GGML on next launch).
Adds an is_success() check in the non-resume branch: any non-2xx
returns KonError::DownloadFailed with the HTTP status in the
message, and (importantly) we return before File::create so no
.part file is left behind.
Test: spawn_500_server that responds 500 to any request; a fresh
(no .part, no sha256) download must Err with "HTTP 500" and leave
neither .part nor dest on disk.
MAJOR from the 2026-04-22 review (wav.rs:135-145): read_wav used
filter_map(|s| s.ok()) on both integer and float sample iterators, so
any per-sample decode error (truncated payload, corrupted format
descriptor after a partial write) was silently discarded. Callers
received Ok with a short samples vec, losing audio without any signal
to investigate.
Switches to collect::<Result<Vec<f32>>>() with a map that converts
hound's per-sample errors into KonError::AudioDecodeFailed. First
error aborts the read rather than returning a partial vector.
Test fabricates the regression by writing a valid WAV and chopping
the last 10 bytes; the previous filter_map would have returned Ok
with a shortened vec, the new code returns Err.
CRITICAL from the 2026-04-22 code review: RmsVadChunker::flush() was
calling consume_frame() on a zero-padded final frame via `let _ = ...`,
discarding any VadChunk the call emitted. If the padded frame triggered
end-of-utterance (silent tail + padding zeros push past
silence_close_samples) or max_chunk_samples (buffered speech + padding
push past the cap), the emitted chunk was lost; the outer state check
then either returned None or an empty closing chunk.
Changes the VadChunker trait flush signature from Option<VadChunk> to
Vec<VadChunk> so both the mid-flush emission (from consume_frame) and
the closing emission (from emit_active_chunk_and_close) can be
surfaced. Updates RmsVadChunker::flush to collect from both sites
and skip a zero-length closing emit when the hit_max continue variant
already cleared active_chunk.
Two regression tests land alongside:
- flush_preserves_hit_max_chunk_from_padded_final_frame: tight
max_chunk, sub-frame speech tail; pre-fix dropped the chunk, post-fix
emitted samples cover the full active-speech region.
- flush_preserves_end_of_utterance_chunk_from_padded_final_frame:
silent tail near silence_close; padded zero frame closes the
utterance inside consume_frame; pre-fix returned None.
No production callers yet — the VadChunker wiring in live.rs is a
deferred item from A.3. API change is clean within the repo.
Review feedback (MINOR): char::is_whitespace returns false for
zero-width format codepoints (U+200B ZWSP, U+200C ZWNJ, U+200D ZWJ,
U+2060 WORD JOINER, U+FEFF ZWNBSP / BOM). The original normalise
pass let them through to the LLM where they waste tokens without
contributing any natural-language content.
Makes the decision explicit: these chars STRIP entirely rather than
collapse to a space. Collapsing would silently insert a word break
where the source had none ("hello<FEFF>world" → "hello world"
would merge two words into a space-separated pair that the original
author did not intend). Stripping preserves the original token
boundaries and drops the invisible noise.
Three new tests:
- zero_width_format_chars_strip_entirely — exhaustive coverage of
all five handled codepoints.
- zero_width_chars_do_not_break_adjacent_whitespace_collapsing —
"hello <FEFF> world" still collapses to "hello world" (the
strip does not leave behind an artefact that breaks the whitespace
collapse pass).
- leading_bom_is_stripped — a BOM at segment start, the common
artefact pattern when Whisper consumes an encoded file.
New crates/ai-formatting/src/to_plain_text.rs module with one public
function: to_plain_text(&[Segment]) -> String.
Rules the function enforces:
- each segment's text is whitespace-normalised (any run of unicode
whitespace collapses to a single ASCII space, so tabs, newlines,
and NBSPs never reach the LLM),
- empty and whitespace-only segments are dropped,
- remaining segments are joined with a single ASCII space,
- the joined string is normalised again (so a segment ending in a
space followed by one starting in a space does not produce a double
space) and trimmed end-to-end.
pipeline.rs's inline join is replaced with this call. Whisper's
timestamp fields (Segment.start / .end) are carried separately and
never reach the LLM by construction — the "timestamps stripped"
half of brief item #29's acceptance falls out of using Segment.text
alone. The work the module actually adds is whitespace discipline
and the tested boundary (empty input, empty-only input, NBSPs,
pathological whitespace runs, idempotence, double-space at join
boundaries).
Source: Scriberr PR #288 — feeding raw Whisper JSON (with timestamps
and per-segment structure) degraded cleanup quality; plain-text
input raised it back.
Review feedback (MINOR): the original <= 0.0 clamp caught negatives
and zero but not non-finite inputs. Rust's saturating float-to-int
cast turns f64::INFINITY into u64::MAX, which would park the capture
buffer origin beyond any reachable sample index and trim the whole
buffer forever if a future end_secs source ever produces infinity
(clock glitch, overflow upstream, corrupted timestamp in a pass).
Adds is_finite() check. NaN, +infinity, -infinity, and zero all
return 0, which downstream trim_buffer_to_commit_point treats as
no-op. Test covers all three non-finite cases.
Review feedback (CRITICAL): when a chunk hit max_chunk_samples during
continuous speech, emit_active_chunk reset state to Idle. The next
1-2 loud frames of post-split continued speech went into onset_buffer
and were silently cleared if silence arrived before the 3-frame onset
threshold — 50-100ms of user audio lost at every max-chunk boundary
in long-continuous-speech scenarios.
Splits emit_active_chunk into two variants:
- emit_active_chunk_and_close: the existing behaviour. Used for
end-of-utterance closes and end-of-session flush. Truncates trailing
silence, resets state to Idle.
- emit_active_chunk_continue: mid-utterance split on max_chunk. Stays
in State::InSpeech, clears active_chunk for continued accumulation,
advances active_chunk_start by the emitted length so the next
chunk's start_sample is contiguous with this one's end. No
silence-trim (by definition still in speech — end-of-utterance
takes priority).
Adds max_chunk_split_preserves_audio_contiguity test: feeds 17 frames
of continuous speech into a chunker with a 4-frame cap, asserts
(a) chunk[i+1].start_sample == chunk[i].start_sample + chunk[i].samples.len()
across every pair, and (b) the final emitted region reaches the end
of the fed speech with no sample loss.
Review feedback (CRITICAL): LocalAgreement::push could panic with an
index OOB when a later pass arrived shorter than committed_count.
Concrete case: commit [a, b], next pass arrives [a] — lcp_len=1,
new_committed=max(1, 2)=2, then latest[2..] panicked because
latest.len()==1.
A Whisper re-transcription of an overlapping window can legitimately
collapse repeated segments, or the user can stop mid-utterance after
some tokens were already committed, both of which produce this
shape. The committed_count invariant still holds (non-shrinkage) —
it is the slicing that was unsafe.
Clamps every latest[..] slice against latest.len() before indexing.
committed_count stays at new_committed even when the pass is shorter:
non-shrinkage is relative to what we have already emitted, not to
the current pass length. newly_committed and tentative both return
empty when the shorter pass has nothing past the committed prefix.
Adds two regression tests:
- shorter_pass_after_commit_does_not_panic (commit 2, push 1)
- empty_pass_after_commit_does_not_panic (commit 1, push empty)
New streaming::buffer_trim module with two pure helpers:
- sample_index_for_seconds(end_secs, sample_rate) -> u64: converts
LocalAgreement::last_committed_end_secs() into an absolute sample
index. Defensive against negative end_secs (treats as 0) so a future
clock-skewed timestamp cannot wrap to a huge u64.
- trim_buffer_to_commit_point(buffer, buffer_start_sample,
commit_sample_index) -> new_buffer_start_sample: drains the prefix
of the capture buffer that falls below the commit point and returns
the new absolute-index origin.
Edge cases covered by tests:
- commit before buffer start → no drain
- commit equal to buffer start → no drain
- commit inside buffer → drain prefix, advance origin
- commit at buffer end → drain all, origin moves forward
- commit past buffer end → drain all, origin parks at commit (rare
edge after a committer reset)
- sample_index_for_seconds rounds nearest, negatives clamp to 0
- integration with LocalAgreement::last_committed_end_secs
trim_bounds_buffer_over_long_session is the acceptance fixture for
the ufal #120/#102 failure mode: 100 cycles of 16_000 captured samples
with a 200-sample tentative tail per cycle, and the buffer stays below
2× the tentative envelope instead of growing to 1.6M samples.
Integration into src-tauri/src/commands/live.rs deferred to the
dogfood session that wires VadChunker and LocalAgreement end-to-end —
the trim is a one-line replacement at the maybe_dispatch_chunk drain
site once the committer is feeding it commit points.
