Replace two-step docuvision OCR + LLM structuring pipeline with a
single multimodal VLM call. The bartowski Qwen2-VL-7B-Instruct Q5_K_M
GGUF is served by cf-text (llama.cpp) and accepts image_url content
blocks identical to the OpenAI vision API format.
Removes docuvision dependency for recipe scanning; the addict-missing /
DeepseekVLV2Processor-missing cf-docuvision error no longer blocks scans.
Receipt OCR (kiwi.ocr task) still routes to cf-docuvision separately.
POST /recipes/scan/stream emits live status events while cf-docuvision
allocates and processes, replacing the static spinner with phase-aware labels:
allocating -> scanning -> structuring -> done|error
Uses asyncio.Queue bridge to route progress callbacks from the sync scanner
thread to the async SSE generator. Frontend updated to consume the stream via
fetch + ReadableStream (EventSource does not support POST multipart).
Closes kiwi#136 (companion to the docuvision routing fix).
Two-step pipeline: task_allocate("kiwi", "recipe_scan", service_hint="cf-docuvision")
acquires a docuvision allocation, calls /extract per image to get OCR text, then
LLMRouter structures the combined OCR output into recipe JSON via the text
extraction prompt.
Also fixes DocuvisionClient bugs:
- POST field was "image" (ignored by Pydantic) — should be "image_b64"
- Response read "text" key — docuvision returns "raw_text"
- Add hint parameter (use "text" for recipe cards, dense prose)
- Configurable timeout (default 120s; docuvision lazy-loads model on first request)
Replaces single-path cf-orch allocation with a three-tier strategy:
tier 1 task_allocate() (coordinator-driven), tier 2 direct CFOrchClient.allocate()
(TaskNotRegistered fallback), tier 3 local LLMRouter. Module-level imports for
CFOrchClient and LLMRouter make all three paths patchable in tests without
import caching issues.
Full ActivityPub implementation wired to cf-core.activitypub module:
Endpoints (root-level, not under /api/v1):
GET /.well-known/webfinger — WebFinger JRD (AP_ENABLED only)
GET /ap/actor — Instance actor document
POST /ap/actor/inbox — Incoming Follow/Undo (dedup + Accept dispatch)
GET /ap/outbox — OrderedCollection of community posts
GET /ap/posts/{slug} — Individual AP Note
GET /ap/followers — Follower count collection
GET /ap/following — Empty following collection
Mastodon OAuth (under /api/v1/social/mastodon/):
POST /connect — Dynamic app registration + OAuth flow start
GET /callback — Code exchange + token storage (Fernet-encrypted)
DELETE /disconnect — Token revocation
GET /status — Connection status
Config: AP_ENABLED, AP_HOST, AP_KEY_PATH, AP_TOKEN_ENCRYPTION_KEY
Migration 042: ap_followers, ap_deliveries, ap_received, mastodon_tokens tables
Key manager: auto-generates RSA-2048 keypair on first boot if AP_ENABLED
Delivery service: deliver_to_followers() with 3-retry exponential backoff + DB log
Post publish: background fan-out to AP followers + Mastodon when opted-in
All AP endpoints gracefully degrade (404) when AP_ENABLED=false.
Three-layer dedup check before community post submission:
- L1: title ILIKE search against existing posts in community DB
- L2: Jaccard ingredient overlap using local corpus (≥0.70 very_similar, ≥0.35 somewhat_similar)
- L3: similar_to_ref FK — user can explicitly mark post as variation of existing
New endpoint: POST /api/v1/community/check-similar (gracefully no-ops if community DB absent)
New service: app/services/community/dedup.py — jaccard(), similarity_tier(), build_similar_post_result()
Both publish modals (plan + outcome) now check similarity before submit; user can proceed as-is,
mark as variation, or cancel. similar_to_ref passed in final publish payload.
Free tier: keyword extraction + FTS ingredient search + title probe search.
Paid tier / BYOK: same search, then LLM synthesis of a conversational answer
(8s timeout so an unresponsive model degrades gracefully to recipe list only).
- AskRequest / AskRecipeHit / AskResponse schemas in recipe.py
- _extract_ask_keywords(): tokenize question, strip stopwords
- _ask_in_thread(): two-pronged search (ingredient FTS + title LIKE)
merges by ID, computes pantry match_pct when pantry_items provided
- Endpoint registered before /{recipe_id} to avoid integer coercion on /ask
- LLM synthesis gated to paid/premium/ultra only (not "local" dev tier)
Closes#134 (backend)
Previously, get_or_create_product was only called when auto_add was true,
so scan responses with auto_add=false returned no product details. Now the
DB lookup always runs when product_info is available; inventory insertion
is still conditional on auto_add_to_inventory. Fixes preview-only barcode
scans returning empty product fields.
Two-phase streaming architecture:
Phase 1 (sync thread): IngredientClassifier builds element profiles +
gap list from SQLite — thread-safe, no async context needed
Phase 2 (async): LLMRecipeGenerator.stream_generate() yields tokens via
cf-orch warm vllm (existing /stream-token path) or AsyncOpenAI against
Ollama if the coordinator is unavailable
Backend (app/services/recipe/llm_recipe.py):
- stream_generate() async generator; _try_alloc_for_stream() sync helper
- _stream_openai_compat() static method handles __auto__ model resolution
- LLMRecipeGenerator(None) is safe for streaming (store not used)
Endpoint (app/api/endpoints/recipes.py):
- ?stream=true on POST /recipes/suggest returns StreamingResponse
- X-Accel-Buffering: no prevents nginx buffering without nginx.conf edits
Frontend (api.ts, recipes.ts, RecipesView.vue):
- suggestRecipeStream() uses fetch + ReadableStream (POST; EventSource
only supports GET)
- streamSuggest() action in recipes store builds request internally
- RecipesView.streamRecipe() silently falls back to native SSE when
cf-orch token fetch fails rather than surfacing an error
Exposes four read-only tools to Claude Code:
kiwi_query_corpus — parameterised SELECT against kiwi.db (200-row cap)
kiwi_count_fts — FTS5 MATCH hit count for keyword coverage audits
kiwi_sample_tags — tag frequency distribution by prefix
kiwi_browse_preview — first-page results from the live browse API
DB opened in SQLite URI read-only mode (mode=ro); any write statement is
rejected at the driver level. Configure via KIWI_DB_PATH and KIWI_API_URL
env vars (see module docstring for settings.json snippet).
Adds max_active_min request field and backend filter. Active time uses
parse_time_effort().active_min (passive waits excluded). Recipes with
no parsed active time signal are not excluded (avoid hiding unlabelled
results). Total and active limits are AND'd when both set.
UI: two pill rows — "Hands-on time" (15/30/45/1hr) and "Total time"
(30m/1hr/90m/2hr/3hr/4+hr). Replaces single row capped at 90 min.
Recipe cards were rendering full directions, all nutrition chips,
prep notes, swap candidates, and grocery links inline in the grid —
making each card tall enough to push the second row below the fold at
3-column widths. Cards now show title, match/complexity/time badges,
up to 4 pantry ingredient chips, missing count, and calorie hint.
Full detail remains in RecipeDetailPanel on "Make this".
ElementClassifier.classify_batch() was issuing N separate DB queries
(one per pantry item). Replaced with a single WHERE name IN (...)
query + heuristic fallback for misses — same result, one round-trip.
Adds _MEAL_SIGNALS table to tag_inferrer with title-only matching for:
meal:Breakfast — pancakes, waffles, frittata, oatmeal, granola, etc.
meal:Dessert — cake, cookie, brownie, pudding, ice cream, tart, etc.
meal:Snack — dip, chips, popcorn, nachos, energy balls, etc.
meal:Beverage — smoothie, cocktail, juice, lemonade, etc.
meal:Lunch — sandwich, wrap, panini, grilled cheese, etc.
meal:Bread — bread, sourdough, focaccia, dinner roll, etc.
Uses word-boundary + optional-plural regex (\bWORD(?:s|es)?\b) so:
- "pancakes" matches the "pancake" pattern but "pancake" != "cake"
- "tartare" does not match "tart" (no word boundary after tart in tartare)
- "dipping" does not match "dip" (extra chars prevent boundary)
Title-only matching (not ingredient text) avoids false positives from
ingredient names like "cake flour" or "sandwich bread".
Estimated browse impact after backfill (--force on 3.19M recipes):
Breakfast: 43 → ~70k
Dessert: 372 → ~350k (real desserts, not flavor:Sweet)
Snack: 57 → ~60k
Beverage: 43 → ~36k
Lunch: 69 → ~26k
- Dinner: replace non-matching text keywords with main:X protein inferred tags (0 -> 815k results)
- All meal_type categories: add meal:X structured tag phrases
- Dietary: switch to dietary:X-only phrases; bare text keywords matched can_be:X
tags (nearly all recipes), inflating counts to 1.3M+ falsely
- Cuisine: add cuisine:X structured tag phrases to Italian, Mexican, Asian,
Indian, Mediterranean, American, BBQ, European, Latin American
- Side Dish: use main:Vegetables + main:Grains as proxy (no meal:Side Dish tag exists)
- Dessert: remove 'sweet' keyword (matched flavor:Sweet on all recipes)
- New dietary categories: Low-Sodium, Paleo
Closes#122. Partial progress on #123.
Follow-up: #125 (expand meal: tag inferrer coverage)
New feature: photograph a recipe card, cookbook page, or handwritten
note and have it extracted into a structured, editable recipe.
Backend:
- POST /recipes/scan: accept 1-4 photos, run VLM extraction, return
structured JSON for review (not auto-saved)
- POST /recipes/scan/save: persist a reviewed/edited recipe
- GET/DELETE /recipes/user: user-created recipe CRUD
- Vision backend priority: cf-orch -> local Qwen2.5-VL -> Anthropic BYOK
- 503 with clear config hint when no vision backend available
- Multi-photo support: facing pages (ingredients/directions) sent together
- Pantry cross-reference: marks which ingredients are already on hand
- migration 041: user_recipes table (title, servings, cook_time, steps,
ingredients JSON, source, pantry_match_pct)
- Tier gate: recipe_scan -> paid, BYOK-unlockable
Frontend:
- "Scan" button in the Recipes tab bar (camera icon)
- RecipeScanModal: upload step (drag-drop + file picker, up to 4 photos,
live previews), processing step (spinner), review/edit step (all
fields inline-editable before save), pantry match badge, warning banner
for low-confidence or incomplete scans
Tests: 35 new tests (23 unit + 12 API), 404 total passing
Style classifier (kiwi#27):
- app/services/recipe/style_classifier.py: LLM prompt with curated vocab,
cf-orch/LLMRouter fallback, JSON + regex tag extraction
- POST /recipes/saved/{recipe_id}/classify-style: Paid/BYOK tier gate,
fetches recipe from corpus, returns {suggested_tags:[...]}
- SaveRecipeModal.vue: "Suggest tags" button with loading state; merges
LLM suggestions into existing tags without overwriting user's choices
- 403/empty list silently ignored — button is a no-op when tier not met
Cooked leftovers shelf-life (kiwi#112):
- app/services/leftovers_predictor.py: deterministic FDA/USDA lookup table
with shortest-component-wins for proteins and dish-type override for
assembled dishes; special entries for ceviche (2d, acid != heat),
fermented/cured (kimchi 14d, confit/lardo 7d), soups, rice, pasta, etc.
- POST /recipes/{recipe_id}/leftovers: free tier, no gate
- RecipeDetailPanel.vue: shelf-life section appears after "I cooked this"
with fridge/freeze days, freeze-by advice, per-instance dismiss; calm
framing per no-panic UX policy
- LeftoversResponse Pydantic schema added to recipe.py
- Add TimeEffortProfile + StepAnalysis Pydantic schemas; serialised into
RecipeSuggestion so the frontend receives active/passive/total minutes,
effort label, and detected equipment per suggestion.
- parse_time_effort() now drives max_total_min filter (falls back to step-count
estimate when directions contain no explicit time mentions).
- _PRODUCT_TOKEN_STOPWORDS: strips marketing/packaging words from multi-word
product labels before adding individual ingredient tokens to pantry_set.
"Organic Extra Firm Tofu" → adds "tofu"; improves packaged-food pantry match.
- L1 candidate pool raised to 60 (was 20); min_match_ratio lowered to 0.35
(was 0.60) to keep enough results for plant-based / packaged-food pantries.
- household.py: tighten import to pull HEIMDALL_URL/ADMIN_TOKEN from
services.heimdall_orch (matches refactor in cloud_session.py).
- Migration 039: drop saved_recipes.recipe_id FK (SQLite table rebuild).
The FK referenced main.recipes but corpus lives in an ATTACH'd DB — caused
500 on every POST /recipes/saved in cloud mode.
- _to_summary: row.get("title") or "" to handle corpus JOIN returning NULL
title (e.g. placeholder recipe_id 99999).
- list_collections: return [] for Free tier instead of 403 — prevents
Promise.all in savedStore.load() from aborting the saved-recipes fetch.
- savedStore.load(): switched to Promise.allSettled so a collections failure
never blocks the saved list from populating.
- RecipesView: star indicator now reflects savedStore.isSaved() (server-side
saved state) rather than localStorage bookmarks; changed to <span> since
the star is now read-only visual feedback.
- Removed { immediate: true } from saved-tab watcher — premature bounce to
Build Your Own before onMounted load() completes.
Delegates JWT validation, Heimdall provision/tier-resolve, bypass-IP
handling, and guest session management to circuitforge_core. Kiwi keeps
its own CloudUser (db path, household fields, BYOK flag) and DB helpers.
detect_byok() is now imported from cf-core instead of a local copy.
household_id/is_household_owner/license_key flow through core_user.meta
(cf-core already forwards all Heimdall response extras into meta).
Removes ~217 lines of duplicated auth code.
Note: guest cookie name changes from kiwi_guest_id to cf_guest_id (cf-core
managed). Existing guest sessions get a new UUID on first visit — acceptable
for alpha.
Migrations 037 and 038 rebuild products and inventory_items tables
to add 'visual_capture' as a valid source value, which the label-confirm
endpoint sets when saving user-verified nutrition label data.
Adds 2 schema tests covering the new allowed value.
When a barcode scan finds no product in FDC/OFF, paid-tier users now see a
"Capture label" offer instead of a dead-end "add manually" prompt.
Backend:
- Migration 036: captured_products local cache table (keyed by barcode,
UPSERT on conflict so re-capture refreshes rather than errors)
- store.get_captured_product / save_captured_product (with JSON decode for
ingredient_names and allergens)
- app/services/label_capture.py: wraps cf-core VisionRouter (caption API);
graceful fallback to zero-confidence mock when stub/error; JSON fence
stripping; confidence clamped to [0,1]; KIWI_LABEL_CAPTURE_MOCK=1 for tests
- New schemas: LabelCaptureResponse, LabelConfirmRequest, LabelConfirmResponse
- POST /inventory/scan/label-capture — image to extraction (paid+ gate, 403)
- POST /inventory/scan/label-confirm — save confirmed product + optional
inventory add
- Both scan endpoints now: check captured_products cache before FDC/OFF;
set needs_visual_capture=True for gap products on paid tier; BarcodeScanResult
gains needs_visual_capture field
- visual_label_capture feature gate added to tiers.py (paid)
Tests: 42 new tests (service, store/migration, API endpoints) — 367 total passing
Frontend:
- InventoryList.vue: capturePhase state machine (offer => uploading => reviewing)
- Offer card appears after scan gap (calm UX: no urgency, Discard always visible)
- Review form: pre-populated from extraction; amber label highlights for
unread fields (confidence < 0.7); comma-separated ingredients/allergens
- api.ts: LabelCaptureResult + LabelConfirmRequest types; captureLabelPhoto()
and confirmLabelCapture() API methods
_user_constraints() loads dietary_constraints from user_settings once per
request. All 7 _enrich_item call sites now pass constraints so wine (and
any future alcohol-containing entries) are suppressed for halal/alcohol-free
users at the API response layer.
Adds 10 new secondary use entries and corrects all 8 existing ones.
New: apples/soft, leafy_greens/wilting, tomatoes/soft, cooked_pasta/day-old,
cooked_potatoes/day-old, yogurt/tangy, cream/sour, wine/open,
cooked_beans/day-old, cooked_meat/leftover.
Corrections: milk uses (specific recipes, not 'baking'/'sauces'); dairy uses
expanded; cheese label well-aged→rind-ready with named dishes (minestrone,
ribollita); rice uses (onigiri, arancini, congee); tortillas warning added;
bakery uses and synonyms expanded to named pastries; bananas synonyms
(spotty/brown/black/mushy); rice synonyms (old rice).
New fields on every SECONDARY_WINDOW entry:
- discard_signs: qualitative cues for when the item has gone past its
secondary window (shown in UI alongside uses)
- constraints_exclude: dietary labels that suppress the entry entirely
(wine suppressed for halal/alcohol-free)
ExpirationPredictor.filter_secondary_by_constraints() applies constraint
suppression; _enrich_item() now accepts user_constraints and passes
secondary_discard_signs through to the API response.
Integrates cf-core reranker into the L1/L2 recipe engine. Paid+ tier
gets a BGE cross-encoder pass over the top-20 FTS candidates, scoring
each recipe against the user's full context: pantry state, dietary
constraints, allergies, expiry urgency, style preference, and effort
preference. Free tier keeps the existing overlap sort unchanged.
- New app/services/recipe/reranker.py: build_query, build_candidate_string,
rerank_suggestions with tier gate (_RERANKER_TIERS) and graceful fallback
- rerank_score field added to RecipeSuggestion (None on free tier, float on paid+)
- recipe_engine.py: single call after candidate assembly, before final sort;
hard_day_mode tier grouping preserved as primary sort when reranker active
- Fix pre-existing circular import in app/services/__init__.py (eager import
of ReceiptService triggered store.py → services → receipt_service → store)
- 27 unit tests (mock backend, no model weights) + 2 engine-level tier tests;
325 tests passing, no regressions
- Add max_total_min to RecipeRequest schema and TypeScript interface
- Add _within_time() helper to recipe_engine using parse_time_effort()
with graceful degradation (empty directions or no signals -> pass)
- Wire max_total_min filter into suggest() loop after max_time_min
- Add time_first_layout to allowed settings keys
- Add timeFirstLayout ref to settings store (preserves sensoryPreferences)
- Add maxTotalMin ref to recipes store, wired into _buildRequest()
- Add time bucket selector UI (15/30/45/60/90 min) in RecipesView
Find tab, gated by timeFirstLayout != 'normal'
- Add time-first layout selector section in SettingsView
- Add 5 _within_time unit tests and 2 settings key tests
- Add app/services/recipe/time_effort.py: parse_time_effort(), TimeEffortProfile,
StepAnalysis dataclasses; two-branch regex for time ranges and single values;
whole-word passive keyword detection; 480 min/step cap; 1825 day global cap
- Add directions to browse_recipes and _browse_by_match SELECT queries in store.py
- Enrich browse and detail endpoints with active_min/passive_min/time_effort fields
- Add StepAnalysis, TimeEffortProfile TS interfaces to api.ts
- RecipeBrowserPanel: split pill badge showing active/passive time
- RecipeDetailPanel: collapsible ingredients summary, effort cards (Active/Hands-off/Total),
equipment chips, annotated step list with Active/Wait badges and passive hints
- 45 new tests (40 unit + 5 API); 215 total passing
browse_counts_cache.py: after FTS counts, _merge_community_tag_counts() queries
accepted tags (upvotes>=2) grouped by (domain,category,subcategory) and adds
distinct recipe_id counts to the cached keyword-set totals. Skips silently
when community Postgres is unavailable.
store.py: fetch_recipes_by_ids() fetches corpus recipes by explicit ID list,
used by the FTS fallback when a subcategory returns zero FTS results.
recipes.py (browse endpoint): when FTS total==0 for a subcategory, queries
community store for accepted tag IDs and serves those recipes directly.
Sets community_tagged=True in the response so the UI can surface context.
Refs kiwi#118.
GET /recipes/community-tags/{recipe_id} — all tags for a recipe
POST /recipes/community-tags — submit tag (requires pseudonym)
POST /recipes/community-tags/{id}/upvote — vote on a tag
Validates (domain, category, subcategory) against DOMAINS taxonomy before
accepting. Returns 409 on duplicate submission or double-vote. Fails soft
(503) when community Postgres is unavailable so the browse path is unaffected.
Refs kiwi#118.
food.com rarely tags BBQ in its own taxonomy fields, so BBQ recipes were
previously untagged. Added content-derived signals (brisket, pulled pork,
dry rub, regional styles) so infer_recipe_tags.py correctly tags them
as cuisine:BBQ. Companion to the browser_domains.py BBQ keyword expansion.
Switches recipe generation service type from 'cf-text' to 'vllm' so the
coordinator can route to quantized small models (Qwen2.5-3B, Phi-4-mini)
rather than the full text backend. Passes CF_APP_NAME for per-product
VRAM/request analytics in the coordinator dashboard.
- llm_recipe.py: _SERVICE_TYPE = 'vllm'; _MODEL_CANDIDATES list; passes
model_candidates and pipeline= to CFOrchClient.allocate()
- compose.cloud.yml: CF_APP_NAME=kiwi env var for coordinator attribution
Shopping links previously hardcoded to US storefronts. Users in other regions
got broken Amazon Fresh and Instacart links. Now locale is stored as a user
setting and passed to GroceryLinkBuilder at request time.
- locale_config.py: per-locale Amazon domain/dept config (already existed)
- grocery_links.py: GroceryLinkBuilder accepts locale=; routes Instacart to .ca
for Canada, uses amazon_domain per locale, Instacart/Walmart US/CA only
- settings.py: adds 'shopping_locale' to allowed settings keys
- shopping.py: reads locale from user's stored setting on all list/add/update paths
- SettingsView.vue: Shopping Region selector (NA, Europe, APAC, LATAM)
- stores/settings.ts: shoppingLocale reactive state, saves via settings API
Users often have ingredients they want to avoid today (out of stock, not feeling it)
that aren't true allergies. The new 'Not today' filter lets them exclude specific
ingredients per session without permanently modifying their allergy list.
- recipe.py schema: exclude_ingredients field (list[str], default [])
- recipe_engine.py: filters corpus results when any ingredient is in exclude_set
- llm_recipe.py: injects exclusions into both prompt templates so LLM-generated
recipes respect the constraint at generation time
- RecipesView.vue: tag-chip UI with Enter/comma input, removes on × click
- stores/recipes.ts: excludeIngredients reactive list (not persisted to localStorage)
Adds 'Best match' sort button to the recipe browser. When selected, recipes are
ordered by the fraction of their ingredients that are in the user's pantry.
- store.py: _browse_by_match() pushes match_pct computation into SQL via json_each()
so ORDER BY can sort the full result set before LIMIT/OFFSET pagination
- recipes.py: extends sort pattern validation to accept 'match'; falls back to
default when no pantry_items provided
- RecipeBrowserPanel.vue: adds 'Best match' button (disabled when pantry empty);
watcher auto-engages match sort when pantry goes from empty to non-empty
Multiple concurrent users browsing the 3.2M recipe corpus would cause FTS5 page
cache contention and slow per-request queries. Solution: pre-compute counts for
all category/subcategory keyword sets into a small SQLite cache.
- browse_counts_cache.py: refresh(), load_into_memory(), is_stale() helpers
- config.py: BROWSE_COUNTS_PATH setting (default DATA_DIR/browse_counts.db)
- main.py: warms in-memory cache on startup; runs nightly refresh task every 24h
- infer_recipe_tags.py: auto-refreshes cache after a successful tag run so the
app picks up updated FTS counts without a restart
Top-level category keywords were too narrow, missing common food.com corpus terms
like 'barbecue', 'smoky', 'charcoal'. Subcategory terms also expanded to cover
broader corpus vocabulary so FTS counts register hits across more recipes.
Paid+ local users with circuitforge_orch installed now get the coordinator-
aware scheduler automatically — no env var needed. The coordinator's
allocation queue already prefers the local GPU first, so latency stays low.
Priority: USE_ORCH_SCHEDULER env override > CLOUD_MODE > cf-orch importable.
Free-tier local users without cf-orch installed get LocalScheduler as before.
USE_ORCH_SCHEDULER=false can force LocalScheduler even when cf-orch is present.
Switches to OrchestratedScheduler in cloud mode so concurrent recipe_llm
jobs fan out across all registered cf-orch GPU nodes instead of serializing
on one. Under load this eliminates poll timeouts from queue backup.
USE_ORCH_SCHEDULER env var gives explicit control independent of CLOUD_MODE:
unset follow CLOUD_MODE (cloud=orch, local=local)
true OrchestratedScheduler always (e.g. multi-GPU local rig)
false LocalScheduler always (e.g. cloud single-GPU dev instance)
ImportError fallback: if circuitforge_orch is not installed and orch is
requested, logs a warning and falls back to LocalScheduler gracefully.
Adds the recipe_jobs table and background task pipeline for level 3/4
recipe generation. POST ?async=true returns 202 with job_id; clients
poll GET /recipes/jobs/{job_id} until status=done.
Key fix: _enqueue_recipe_job now calls scheduler.enqueue() after
insert_task() to wake the in-memory work queue immediately. Without
this, tasks sat in 'queued' until the scheduler's 30s idle cycle or
an API restart triggered _load_queued_tasks().
- Migration 034: recipe_jobs table (job_id, user_id, status, request,
result, error) with indexes on job_id and user_id/created_at
- Store: create/get/update_running/complete/fail recipe job methods
- runner.py: recipe_llm task type + _run_recipe_llm handler; MUST
call fail_recipe_job() before re-raising so status stays consistent
- CLOUD_MODE guard: falls back to sync generation (scheduler only
polls shared settings DB, not per-user DBs)
- L4 wildcard is covered by the same req.level in (3, 4) dispatch
When the coordinator returns 429 (all nodes at max_concurrent limit), the previous
code fell back to LLMRouter which is also overloaded at high concurrency. This
caused the request to hang for ~60s before nginx returned a 504.
Now: detect 429/max_concurrent in the RuntimeError message and return "" immediately
so the caller gets an empty RecipeResult (graceful degradation) rather than a timeout.
Adds minimal sort/search to the recipe browser for cognitive access diversity —
linear scanners, alphabet browsers, and keyword diggers each get a different
way in without duplicating the full search tab.
- browse_recipes: q (LIKE title filter) + sort (default/alpha/alpha_desc)
- API endpoint: q/sort query params with validation
- Frontend: debounced search input (350ms) + sort pills (Default/A→Z/Z→A)
- Search and sort reset on domain/category change
- _all path supports q+sort; keyword-FTS path adds AND filter on top
