Auto-detect SSL vs STARTTLS based on port number instead of always
using SMTP + starttls(), which times out on port 465 (implicit SSL).
Authored by: Jack Levy
vote_fetcher was missing from Celery's include list (task not registered with
workers) and had no beat schedule — votes only fetched on-demand when a user
visited a bill's votes page. Stanced bills (pocket_veto/pocket_boost) never had
votes fetched, leaving the alignment page blank.
Add fetch_votes_for_stanced_bills nightly task (4:30 AM UTC) that queues
fetch_bill_votes for every bill any user has stanced but has no stored votes.
Register vote_fetcher in the include list and add it to the polling queue route.
Authored by: Jack Levy
Quoteless unlabeled points (old-format briefs with no citation system) were
being auto-labeled via raw SQL after db.get() loaded them into the session
identity map. SQLAlchemy's commit-time flush could re-emit the ORM object's
cached state, silently overwriting the raw UPDATE.
Fix: run a single bulk SQL UPDATE for all matching rows before any ORM objects
are loaded into the session. The commit is then a clean single-statement
transaction with nothing to interfere. LLM classification of quoted points
continues in a separate pass with normal flag_modified + commit.
Authored by: Jack Levy
Dashboard _get_trending() was querying scores within 1 day only — if the
nightly trend task hadn't run (e.g. worker restarted mid-run), the trending
section returned empty. Now falls back through 1→3→7→30 day windows so
stale scores always surface something.
Trend scorer now wraps per-bill scoring in try/except so a single bad
newsapi/gnews call can't abort the entire 1600-bill run.
Authored by: Jack Levy
- Migration 0019: email_unsubscribe_token column on users (unique, indexed)
- Token auto-generated on first email address save (same pattern as RSS token)
- GET /api/notifications/unsubscribe/{token} — no auth required, sets
email_enabled=False and returns a branded HTML confirmation page
- List-Unsubscribe + List-Unsubscribe-Post headers on every email
(improves deliverability; enables one-click unsubscribe in Gmail/Outlook)
- Unsubscribe link appended to email body plain text
Authored by: Jack Levy
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add email as a second notification channel alongside ntfy:
- Tabbed channel selector: ntfy | Email | Telegram (coming soon) | Discord (coming soon)
- Active channel shown with green status dot on tab
- Email tab: address input, Save & Enable, Test, Disable — same UX pattern as ntfy
- Backend: SMTP config in settings (SMTP_HOST/PORT/USER/PASSWORD/FROM/STARTTLS)
- Dispatcher: _send_email() helper wired into dispatch_notifications
- POST /api/notifications/test/email endpoint with descriptive error messages
- Email fires in same window as ntfy (respects quiet hours / digest hold)
- Telegram and Discord tabs show coming-soon banners with planned feature description
- .env.example documents all SMTP settings with provider examples
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Congress.gov occasionally returns the same member twice on a single page
with different sponsorship dates (observed: Sen. Warnock on 119-s-1383).
The DB uniqueness check didn't catch this because the first insert hadn't
been committed yet when processing the duplicate row, causing a
UniqueViolation. Fix adds an `inserted_this_run` set to skip bioguide_ids
already added in the current fetch loop.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Member Effectiveness Score
- New BillCosponsor table (migration 0018) with per-bill co-sponsor
party data required for the bipartisan multiplier
- bill_category column on Bill (substantive | commemorative | administrative)
set by a cheap one-shot LLM call after each brief is generated
- effectiveness_score / percentile / tier columns on Member
- New bill_classifier.py worker with 5 tasks:
classify_bill_category — triggered from llm_processor after brief
fetch_bill_cosponsors — triggered from congress_poller on new bill
calculate_effectiveness_scores — nightly at 5 AM UTC
backfill_bill_categories / backfill_all_bill_cosponsors — one-time
- Scoring: distance-traveled pts × bipartisan (1.5×) × substance (0.1×
for commemorative) × leadership (1.2× for committee chairs)
- Percentile normalised within (seniority tier × party) buckets
- Effectiveness card on member detail page with colour-coded bar
- Admin panel: 3 new backfill/calculate controls in Maintenance section
Representation Alignment View
- New GET /api/alignment endpoint: cross-references user's stanced bill
follows (pocket_veto/pocket_boost) with followed members' vote positions
- Efficient bulk queries — no N+1 loops
- New /alignment page with ranked member list and alignment bars
- Alignment added to sidebar nav (auth-required)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Submit up to 1000 unbriefed documents to the provider Batch API in one
shot instead of individual synchronous LLM calls. Results are polled
every 30 minutes via a new Celery beat task and imported automatically.
- New worker: llm_batch_processor.py
- submit_llm_batch: guards against duplicate batches, builds JSONL
(OpenAI) or request list (Anthropic), stores state in AppSetting
- poll_llm_batch_results: checks batch status, imports completed
results with idempotency, emits notifications + triggers news fetch
- celery_app: register worker, route to llm queue, beat every 30 min
- admin API: POST /submit-llm-batch + GET /llm-batch-status endpoints
- Frontend: submitLlmBatch + getLlmBatchStatus in adminAPI; settings
page shows batch control row (openai/anthropic only) with live
progress line while batch is processing
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Nine efficiency improvements across the data pipeline:
1. NewsAPI OR batching (news_service.py + news_fetcher.py)
- Combine up to 4 bills per NewsAPI call using OR query syntax
- NEWSAPI_BATCH_SIZE=4 means ~4× effective daily quota (100→400 bill-fetches)
- fetch_news_for_bill_batch task; fetch_news_for_active_bills queues batches
2. Google News RSS cache (news_service.py)
- 2-hour Redis cache shared between news_fetcher and trend_scorer
- Eliminates duplicate RSS hits when both workers run against same bill
- clear_gnews_cache() admin helper + admin endpoint
3. pytrends keyword batching (trends_service.py + trend_scorer.py)
- Compare up to 5 bills per pytrends call instead of 1
- get_trends_scores_batch() returns scores in original order
- Reduces pytrends calls by ~5× and associated rate-limit risk
4. GovInfo ETags (govinfo_api.py + document_fetcher.py)
- If-None-Match conditional GET; DocumentUnchangedError on HTTP 304
- ETags stored in Redis (30-day TTL) keyed by MD5(url)
- document_fetcher catches DocumentUnchangedError → {"status": "unchanged"}
5. Anthropic prompt caching (llm_service.py)
- cache_control: {type: ephemeral} on system messages in AnthropicProvider
- Caches the ~700-token system prompt server-side; ~50% cost reduction on
repeated calls within the 5-minute cache window
6. Async sponsor fetch (congress_poller.py)
- New fetch_sponsor_for_bill Celery task replaces blocking get_bill_detail()
inline in poll loop
- Bills saved immediately with sponsor_id=None; sponsor linked async
- Removes 0.25s sleep per new bill from poll hot path
7. Skip doc fetch for procedural actions (congress_poller.py)
- _DOC_PRODUCING_CATEGORIES = {vote, committee_report, presidential, ...}
- fetch_bill_documents only enqueued when action is likely to produce
new GovInfo text (saves ~60–70% of unnecessary document fetch attempts)
8. Adaptive poll frequency (congress_poller.py)
- _is_congress_off_hours(): weekends + before 9AM / after 9PM EST
- Skips poll if off-hours AND last poll < 1 hour ago
- Prevents wasteful polling when Congress is not in session
9. Admin panel additions (admin.py + settings/page.tsx + api.ts)
- GET /api/admin/newsapi-quota → remaining calls today
- POST /api/admin/clear-gnews-cache → flush RSS cache
- Settings page shows NewsAPI quota remaining (amber if < 10)
- "Clear Google News Cache" button in Manual Controls
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add 4th "Discovery" tab in Alert Filters for member/topic follow notifications,
with per-source enable toggle, independent event-type filters, and per-entity
mute chips (mute specific members/topics without unfollowing)
- Enrich notification event payloads with follow_mode, matched_member_name,
matched_member_id, and matched_topic so each event knows why it was created
- Dispatcher branches on payload.source for member_follow/topic_follow events,
checking source-level enabled toggle, per-event-type filters, and muted_ids/muted_tags
- Add _build_reason helper; ntfy messages append a "why" line (📌/👤/🏷)
- EventRow in notification history shows a small italic reason line
- Update How It Works: fix stale member/topic paragraph, add Discovery alerts item
Authored-by: Jack Levy
Roll-call votes:
- Migration 0017: bill_votes + member_vote_positions tables
- Fetch vote XML directly from House Clerk / Senate LIS URLs
embedded in bill actions recordedVotes objects
- GET /api/bills/{id}/votes triggers background fetch on first view
- VotePanel on bill detail: yea/nay bar, result badge, followed
member positions with Sen./Rep. title, party badge, and state
Alert filter fix:
- _should_dispatch returns True when alert_filters is None so users
who haven't saved filters still receive all notifications
Authored-By: Jack Levy
The dispatcher was suppressing all referral-tier events (committee
referrals) for neutral-mode users, regardless of whether they
directly followed a bill or just followed a topic. This meant
directly-followed bills like HR 7711 and S 3853 showed ✓ in
Recent Alerts but no ntfy notification was ever fired.
Now only topic-follow referral events are suppressed for neutral
users (topic follows are loose and noisy). Direct bill follows and
member follows always receive referral events.
Authored-By: Jack Levy
ZIP lookup (GET /api/members/by-zip/{zip}):
- Two-step geocoding: Nominatim (ZIP → lat/lng) then Census TIGERweb
Legislative identify (lat/lng → congressional district via GEOID)
- Handles at-large states (AK, DE, MT, ND, SD, VT, WY)
- Added rep_lookup health check to admin External API Health panel
congress_api.py fixes:
- parse_member_from_api: normalize state full name → 2-letter code
(Congress.gov returns "Florida", DB expects "FL")
- parse_member_from_api: read district from top-level data field,
not current_term (district is not inside the term object)
Celery beat: schedule sync_members daily at 1 AM UTC so chamber,
district, and contact info stay current without manual triggering
Members page redesign: photo avatars, party/state/chamber chips,
phone + website links, ZIP lookup form to find your reps
Draft letter improvements: pass rep_name from ZIP lookup so letter
opens with "Dear Representative Franklin," instead of generic salutation;
add has_document filter to bills list endpoint
UX additions: HelpTip component, How It Works page, "How it works"
sidebar nav link, collections page description copy
Authored-By: Jack Levy
Weekly Digest (send_weekly_digest Celery task):
- Runs every Monday 8:30 AM UTC via beat schedule
- Queries all followed bills updated in the past 7 days per user
- Sends low-priority ntfy push (Priority: low, Tags: newspaper,calendar)
- Creates a NotificationEvent (weekly_digest type) for RSS feed visibility
- Admin can trigger immediately via POST /api/admin/trigger-weekly-digest
- Manual Controls panel now includes "Send Weekly Digest" button
Local-time quiet hours:
- Browser auto-detects IANA timezone via Intl.DateTimeFormat().resolvedOptions().timeZone
- Timezone saved to notification_prefs alongside quiet_hours_start/end on Save
- Dispatcher converts UTC → user's local time (zoneinfo stdlib) before hour comparison
- Falls back to UTC if timezone absent or unrecognised
- Quiet hours UI: 12-hour AM/PM selectors, shows detected timezone as hint
- Clearing quiet hours also clears stored timezone
Co-Authored-By: Jack Levy
Test notification:
- Click header -> {PUBLIC_URL}/notifications so tapping the test opens the app
Real bill alerts (dispatcher):
- Title reformatted: "New Bill Text: HR 1234" (event type + bill identifier)
- Body: bill full name on first line, AI summary below (300 chars)
- Tags updated per event type (page, memo, siren) instead of generic scroll
- Click header was already set from bill_url; no change needed there
Authored-By: Jack Levy
Notifications:
- New /notifications page accessible to all users (ntfy + RSS config)
- ntfy now supports no-auth, Bearer token, and HTTP Basic auth (for ACL-protected self-hosted servers)
- RSS enabled/disabled independently of ntfy; token auto-generated on first GET
- Notification settings removed from admin-only Settings page; replaced with link card
- Sidebar adds Notifications nav link for all users
- notification_dispatcher.py: fan-out now marks RSS events dispatched independently
Action history:
- Migration 0012: deduplicates existing bill_actions rows and adds UNIQUE(bill_id, action_date, action_text)
- congress_poller.py: replaces existence-check inserts with ON CONFLICT DO NOTHING (race-condition safe)
- Added backfill_all_bill_actions task (no date filter) + admin endpoint POST /backfill-all-actions
Authored-By: Jack Levy
- Fetch bill actions from Congress.gov and populate the action timeline
- Add nightly batch task and beat schedule for active bill actions
- Add admin reprocess endpoint for per-bill debugging
- Add BriefPanel with "What Changed" view and version history
- Add External API Health section with per-source latency testing
- Redesign Manual Controls as health panel with status dots and descriptions
- Add Resume Analysis task for stalled LLM jobs
- Add Backfill Dates & Links task for bills with null metadata
- Fix LLM provider/model DB overrides being ignored (env vars used instead)
- Fix Gemini 404: gemini-1.5-pro deprecated → gemini-2.0-flash
- Fix Anthropic models list: use REST API directly (SDK too old for .models)
- Replace test-LLM full analysis with lightweight ping (max_tokens=20)
- Add has_document field to BillDetail; show "No bill text published" state
- Fix "Introduced: —" showing for bills with null introduced_date
- Add bills_missing_sponsor and bills_missing_metadata to admin stats
- Add GovInfo health check using /collections endpoint (fixes 500 from /packages)
Authored-By: Jack Levy
- get_llm_provider() now accepts provider + model args so DB overrides
propagate through to all provider constructors (was always reading
env vars, ignoring the admin UI selection)
- /test-llm replaced with lightweight ping (max_tokens=20) instead of
running a full bill analysis; shows model name + reply, no truncation
- /api/settings/llm-models endpoint fetches available models live from
each provider's API (OpenAI, Anthropic REST, Gemini, Ollama)
- Admin UI model picker dynamically populated from provider API;
falls back to manual text input on error; Custom model name option kept
- Default Gemini model updated: gemini-1.5-pro → gemini-2.0-flash
Co-Authored-By: Jack Levy
Backend:
- Add fetch_bill_actions task with pagination and idempotent upsert
- Add fetch_actions_for_active_bills nightly batch (4 AM UTC beat schedule)
- Wire fetch_bill_actions into new-bill creation and _update_bill_if_changed
- Add backfill_brief_citations task: detects pre-citation briefs by JSONB
type check, deletes them, re-queues LLM processing against stored text
(LLM calls only — zero Congress.gov or GovInfo calls)
- Add admin endpoints: POST /bills/{id}/reprocess, /backfill-citations,
/trigger-fetch-actions; add uncited_briefs count to /stats
Frontend:
- New BriefPanel component: wraps AIBriefCard, adds amber "What Changed"
badge for amendment briefs and collapsible version history with
inline brief expansion
- Swap AIBriefCard for BriefPanel on bill detail page
- Admin panel: Backfill Citations + Fetch Bill Actions buttons; amber
warning in stats when uncited briefs remain
- Add feature roadmap document with phased plan through Phase 5
Co-Authored-By: Jack Levy
- Drop global unique constraint on news_articles.url; replace with
(bill_id, url) so the same article can appear for multiple bills
- news_fetcher dedup now scoped to bill_id instead of global URL
- Bill detail endpoint triggers a background news fetch when no
articles are stored, so gnews articles surface on next load
Migration 0009.
Co-Authored-By: Jack Levy
Adds Google Trends, NewsAPI, and Google News RSS scoring for members,
mirroring the existing bill interest pipeline. Member profiles now show
a Public Interest chart (with signal breakdown) and a Related News panel.
Key changes:
- New member_trend_scores + member_news_articles tables (migration 0008)
- fetch_gnews_articles() added to news_service for unlimited RSS article storage
- Bill news fetcher now combines NewsAPI + Google News RSS (more coverage)
- New member_interest Celery worker with scheduled news + trend tasks
- GET /members/{id}/trend and /news API endpoints
- TrendChart redesigned with signal breakdown badges and bar+line combo chart
- NewsPanel accepts generic article shape (bills and members)
Co-Authored-By: Jack Levy
- Poller now fetches bill detail on insert to get sponsor (list endpoint
has no sponsor data)
- Add backfill_sponsor_ids task + admin endpoint + UI button to fix the
1,616 existing bills with NULL sponsor_id
- Member name search now matches both "Last, First" and "First Last"
using split_part() on the stored name column; same fix applied to
global search
- Load Bill.sponsor relationship eagerly in get_member_bills to prevent
MissingGreenlet error during Pydantic serialization
- Remove .trim() on search onChange so spaces can be typed
Authored-By: Jack Levy
LLM prompts updated to output {text, citation, quote} objects for every
key_point and risk. govinfo_url stored on BillBrief (migration 0006) so
the frontend can link directly to the source document without an extra
query. AIBriefCard renders § citation chips that expand inline to show
the verbatim quote and a View source → GovInfo link. Old plain-string
briefs continue to render unchanged.
Authored-By: Jack Levy
- Drop hres/sres/hconres/sconres (simple/concurrent resolutions) from poller;
only track hr, s, hjres, sjres (legislation that can become law)
- On first run, seed from 60 days back instead of full congress history,
keeping the bill count to ~1,600 instead of 13,000+
Authored-By: Jack Levy
