feat(phase2): fact/inference labeling, change-driven alerts, admin cleanup

- Add label: cited_fact | inference to LLM brief schema (all 4 providers)
- Inferred badge in AIBriefCard for inference-labeled points
- backfill_brief_labels Celery task: classifies existing cited points in-place
- POST /api/admin/backfill-labels + unlabeled_briefs stat counter
- Expand milestone keywords: markup, conference
- Add is_referral_action() for committee referrals (referred to)
- Two-tier milestone notifications: progress tier (all follow modes) and
  referral tier (pocket_veto/boost only, neutral suppressed)
- Topic followers now receive bill_updated milestone notifications via
  latest brief topic_tags lookup in _update_bill_if_changed()
- Admin Manual Controls: collapsible Maintenance section for backfill tasks
- Update ARCHITECTURE.md and roadmap for Phase 2 completion

Co-Authored-By: Jack Levy

ARCHITECTURE.md

@@ -238,9 +238,11 @@ Indexes: `bill_id`, `topic_tags` (GIN for JSONB containment queries)
 {
   "text": "The bill allocates $50B for defense",
   "citation": "Section 301(a)(2)",
-  "quote": "There is hereby appropriated for fiscal year 2026, $50,000,000,000 for the Department of Defense..."
+  "quote": "There is hereby appropriated for fiscal year 2026, $50,000,000,000 for the Department of Defense...",
+  "label": "cited_fact"
 }
 ```
+`label` is `"cited_fact"` when the claim is explicitly stated in the quoted text, or `"inference"` when it is an analytical interpretation. Old briefs without this field render without a badge (backward compatible).
 
 ---
 
@@ -324,6 +326,7 @@ News articles correlated to a specific member of Congress.
 | user_id | int (FK → users, CASCADE) | |
 | follow_type | varchar | `bill`, `member`, `topic` |
 | follow_value | varchar | bill_id, bioguide_id, or topic name |
+| follow_mode | varchar | `neutral` \| `pocket_veto` \| `pocket_boost` (default `neutral`) |
 | created_at | timestamptz | |
 
 Unique constraint: `(user_id, follow_type, follow_value)`
@@ -397,12 +400,13 @@ Stores notification events for dispatching to user channels (ntfy, RSS).
 | id | int (PK) | |
 | user_id | int (FK → users, CASCADE) | |
 | bill_id | varchar (FK → bills, SET NULL) | nullable |
-| event_type | varchar | e.g. `new_brief`, `bill_updated`, `new_action` |
+| event_type | varchar | `new_document`, `new_amendment`, `bill_updated` |
-| headline | text | Short description for ntfy title |
+| payload | jsonb | `{bill_title, bill_label, brief_summary, bill_url, milestone_tier}` |
-| body | text | Longer description for ntfy message / RSS content |
+| dispatched_at | timestamptz (nullable) | NULL = pending dispatch |
-| dispatched_at | timestamptz (nullable) | NULL = not yet sent |
 | created_at | timestamptz | |
 
+`milestone_tier` in payload: `"progress"` (passed, signed, markup, conference, etc.) or `"referral"` (committee referral). Neutral follows silently skip referral-tier events; pocket_veto and pocket_boost receive them as early warnings.
+
 ---
 
 ## Alembic Migrations
@@ -442,11 +446,12 @@ Auth header: `Authorization: Bearer <jwt>`
 
 | Method | Path | Auth | Description |
 |---|---|---|---|
-| GET | `/` | — | Paginated bill list. Query: `chamber`, `topic`, `sponsor_id`, `q`, `page`, `per_page`, `sort`. |
+| GET | `/` | — | Paginated bill list. Query: `chamber`, `topic`, `sponsor_id`, `q`, `page`, `per_page`, `sort`. Includes `has_document` flag per bill via a single batch query. |
 | GET | `/{bill_id}` | — | Full bill detail with sponsor, actions, briefs, news, trend scores. |
 | GET | `/{bill_id}/actions` | — | Action timeline, newest first. |
 | GET | `/{bill_id}/news` | — | Related news articles, limit 20. |
 | GET | `/{bill_id}/trend` | — | Trend score history. Query: `days` (7–365, default 30). |
+| POST | `/{bill_id}/draft-letter` | — | Generate a constituent letter draft via the configured LLM. Body: `{stance, recipient, tone, selected_points, include_citations, zip_code?}`. Returns `{draft: string}`. ZIP code is used in the prompt only — never stored or logged. |
 
 ### `/api/members`
 
@@ -503,7 +508,7 @@ Auth header: `Authorization: Bearer <jwt>`
 | GET | `/users` | Admin | All users with follow counts. |
 | DELETE | `/users/{id}` | Admin | Delete user (cannot delete self). Cascades follows. |
 | PATCH | `/users/{id}/toggle-admin` | Admin | Promote/demote admin status (cannot change self). |
-| GET | `/stats` | Admin | Pipeline counters: total bills, docs fetched, briefs generated, pending LLM, missing metadata/sponsors/actions, uncited briefs. |
+| GET | `/stats` | Admin | Pipeline counters: total bills, docs fetched, briefs generated, pending LLM, missing metadata/sponsors/actions, uncited briefs, unlabeled briefs (cited objects without a fact/inference label). |
 | GET | `/api-health` | Admin | Test each external API in parallel; returns status + latency for Congress.gov, GovInfo, NewsAPI, Google News. |
 | POST | `/trigger-poll` | Admin | Queue immediate Congress.gov poll. |
 | POST | `/trigger-member-sync` | Admin | Queue member sync. |
@@ -513,6 +518,7 @@ Auth header: `Authorization: Bearer <jwt>`
 | POST | `/backfill-sponsors` | Admin | Queue one-off task to populate `sponsor_id` on bills where it is NULL. |
 | POST | `/backfill-metadata` | Admin | Fill null `introduced_date`, `chamber`, `congress_url` by re-fetching bill detail. |
 | POST | `/backfill-citations` | Admin | Delete pre-citation briefs and re-queue LLM using stored document text. |
+| POST | `/backfill-labels` | Admin | Classify existing cited brief points as `cited_fact` or `inference` in-place — one compact LLM call per brief, no re-generation. |
 | POST | `/resume-analysis` | Admin | Re-queue LLM for docs with no brief; re-queue doc fetch for bills with no doc. |
 | POST | `/bills/{bill_id}/reprocess` | Admin | Queue document + action fetches for a specific bill (debugging). |
 | GET | `/task-status/{task_id}` | Admin | Celery task status and result. |
@@ -570,6 +576,12 @@ Auth header: `Authorization: Bearer <jwt>`
      has no sponsor data), upserts Member, sets bill.sponsor_id
    ↳ New bills → fetch_bill_documents.delay(bill_id)
    ↳ Updated bills → fetch_bill_documents.delay(bill_id) if changed
+   ↳ Updated bills → emit bill_updated notification if action is a milestone:
+       - "progress" tier: passed/failed, signed/vetoed, enacted, markup, conference,
+         reported from committee, placed on calendar, cloture, roll call
+         → all follow types (bill, sponsor, topic) receive notification
+       - "referral" tier: referred to committee
+         → pocket_veto and pocket_boost only; neutral follows silently skip
 
 2. document_fetcher.fetch_bill_documents(bill_id)
    ↳ Gets text versions from Congress.gov (XML preferred, falls back to HTML/PDF)
@@ -641,6 +653,7 @@ All providers implement:
 ```python
 generate_brief(doc_text, bill_metadata) → ReverseBrief
 generate_amendment_brief(new_text, prev_text, bill_metadata) → ReverseBrief
+generate_text(prompt) → str   # free-form text, used by draft letter generator
 ```
 
 ### ReverseBrief Dataclass
@@ -649,8 +662,8 @@ generate_amendment_brief(new_text, prev_text, bill_metadata) → ReverseBrief
 @dataclass
 class ReverseBrief:
     summary: str
-    key_points: list[dict]   # [{text, citation, quote}]
+    key_points: list[dict]   # [{text, citation, quote, label}]
-    risks: list[dict]        # [{text, citation, quote}]
+    risks: list[dict]        # [{text, citation, quote, label}]
     deadlines: list[dict]    # [{date, description}]
     topic_tags: list[str]
     llm_provider: str
@@ -664,16 +677,28 @@ class ReverseBrief:
 {
   "summary": "2-4 paragraph plain-language explanation",
   "key_points": [
-    {"text": "claim", "citation": "Section X(y)", "quote": "verbatim excerpt ≤80 words"}
+    {
+      "text": "claim",
+      "citation": "Section X(y)",
+      "quote": "verbatim excerpt ≤80 words",
+      "label": "cited_fact"
+    }
   ],
   "risks": [
-    {"text": "concern", "citation": "Section X(y)", "quote": "verbatim excerpt ≤80 words"}
+    {
+      "text": "concern",
+      "citation": "Section X(y)",
+      "quote": "verbatim excerpt ≤80 words",
+      "label": "inference"
+    }
   ],
   "deadlines": [{"date": "YYYY-MM-DD or null", "description": "..."}],
   "topic_tags": ["healthcare", "taxation"]
 }
 ```
 
+`label` classification rules baked into the system prompt: `"cited_fact"` if the claim is explicitly stated in the quoted text; `"inference"` if it is an analytical interpretation, projection, or implication not literally stated. The UI shows a neutral "Inferred" badge on inference items only (cited_fact is the clean default).
+
 **Amendment brief prompt** focuses on what changed between document versions.
 
 **Smart truncation:** Bills exceeding the token budget are trimmed — 75% of budget from the start (preamble/purpose), 25% from the end (enforcement/effective dates), with an omission notice in the middle.
@@ -715,6 +740,7 @@ Renders the LLM brief. For cited items (new format), shows a `§ Section X(y)` c
 - Blockquoted verbatim excerpt from the bill
 - "View source →" link to GovInfo (opens in new tab)
 - One chip open at a time per card
+- Inference items show a neutral "Inferred" badge (analytical interpretation, not a literal quote)
 - Old plain-string briefs render without chips (graceful backward compat)
 
 **`ActionTimeline.tsx`**
@@ -729,8 +755,11 @@ Client component wrapping the entire app. Waits for Zustand hydration, then redi
 **`Sidebar.tsx`**
 Navigation with: Home, Bills, Members, Following, Topics, Settings (admin only). Shows current user email + logout button at the bottom. Accepts optional `onClose` prop — when provided (mobile drawer context), renders an X close button in the header and calls `onClose` on every nav link click.
 
+**`DraftLetterPanel.tsx`**
+Collapsible panel rendered below `BriefPanel` on the bill detail page (only when a brief exists). Lets users select up to 3 cited points from the brief, choose stance (YES/NO), tone (short/polite/firm), and optionally enter a ZIP code (not stored). Stance auto-populates from the user's follow mode (`pocket_boost` → YES, `pocket_veto` → NO); clears if they unfollow. Recipient (house/senate) is derived from the bill's chamber. Calls `POST /{bill_id}/draft-letter` and renders the plain-text draft in a readonly textarea with a copy-to-clipboard button.
+
 **`BillCard.tsx`**
-Compact bill preview showing bill ID, title, sponsor with party badge, latest action date, and status.
+Compact bill preview showing bill ID, title, sponsor with party badge, latest action date, status, and a text availability indicator: `Brief` (green, analysis done) / `Pending` (amber, text retrieved but not yet analysed) / `No text` (muted, nothing published on Congress.gov).
 
 **`TrendChart.tsx`**
 Line chart of `composite_score` over time with tooltip breakdown of each data source.
@@ -805,7 +834,7 @@ Separate queues prevent a flood of LLM tasks from blocking time-sensitive pollin
 All LLM providers implement the same interface. Switching providers is a single admin setting change — no code changes, no restart required (the factory reads from DB on each task invocation).
 
 ### JSONB for Flexible Brief Storage
-`key_points`, `risks`, `deadlines`, `topic_tags` are stored as JSONB. This means the schema change from `list[str]` to `list[{text, citation, quote}]` required no migration — only the LLM prompt and application code changed. Old string-format briefs and new cited-object briefs coexist in the same column.
+`key_points`, `risks`, `deadlines`, `topic_tags` are stored as JSONB. This means schema changes (adding `citation`/`quote` in v0.2.0, adding `label` in v0.6.0) required no migrations — only the LLM prompt and application code changed. Old string-format briefs, cited-object briefs without labels, and fully-labelled briefs all coexist in the same column and render correctly at each fidelity level.
 
 ### Redis-backed Beat Schedule (RedBeat)
 The Celery Beat schedule is stored in Redis rather than in memory. This means the beat scheduler can restart without losing schedule state or double-firing tasks.
@@ -915,6 +944,55 @@ Nginx uses `resolver 127.0.0.11 valid=10s` (Docker's internal DNS) so upstream c
 - `introduced_date` shown conditionally (not rendered when null, preventing "Introduced: —")
 - Admin reprocess endpoint: `POST /api/admin/bills/{bill_id}/reprocess`
 
+### v0.5.0 — Follow Modes, Public Browsing & Draft Letter Generator
+
+**Follow Modes:**
+- `follow_mode` column on `follows` table: `neutral | pocket_veto | pocket_boost`
+- `FollowButton` replaced with a mode-selector dropdown (shield/zap/heart icons, descriptions for each mode)
+- `pocket_veto` — alert only on advancement milestones; `pocket_boost` — all changes + action prompts
+- Mode stored per-follow; respected by notification dispatcher
+
+**Public Browsing:**
+- Unauthenticated guests can browse bills, members, topics, and the trending dashboard
+- `AuthModal` gates follow and other interactive actions
+- Sidebar and nav adapt to guest state (no email/logout shown)
+- All public endpoints already auth-free; guard refactored to allow guest reads
+
+**Draft Constituent Letter Generator (email_gen):**
+- `DraftLetterPanel.tsx` — collapsible UI below `BriefPanel` for bills with a brief
+- User selects up to 3 cited points from the brief, picks stance (YES/NO), tone, optional ZIP (not stored)
+- Stance pre-fills from follow mode; clears on unfollow (ref-tracked, not effect-guarded)
+- Recipient derived from bill chamber — no dropdown needed
+- `POST /api/bills/{bill_id}/draft-letter` endpoint: reads LLM provider/model from `AppSetting` (respects Settings page), wraps LLM errors with human-readable messages (quota, rate limit, auth)
+- `generate_text(prompt) → str` added to `LLMProvider` ABC and all four providers
+
+**Bill Text Status Indicators:**
+- `has_document` field added to `BillSchema` (list endpoint) via a single batch `SELECT DISTINCT` — no per-card queries
+- `BillCard` shows: `Brief` (green) / `Pending` (amber) / `No text` (muted) based on brief + document state
+
+### v0.6.0 — Phase 2: Change-driven Alerts & Fact/Inference Labeling
+
+**Change-driven Alerts:**
+- `notification_utils.py` milestone keyword list expanded: added `"markup"` (markup sessions) and `"conference"` (conference committee)
+- New `is_referral_action()` classifier for committee referrals (`"referred to"`)
+- Two-tier notification system: `milestone_tier` field in `NotificationEvent.payload`
+  - `"progress"` — high-signal milestones (passed, signed, markup, etc.): all follow types notified
+  - `"referral"` — committee referral: pocket_veto and pocket_boost notified; neutral silently dropped
+- **Topic followers now receive `bill_updated` milestone notifications** — previously they only received `new_document`/`new_amendment` events. Fixed by querying the bill's latest brief for `topic_tags` inside `_update_bill_if_changed()`
+- All three follow types (bill, sponsor, topic) covered for both tiers
+
+**Fact vs Inference Labeling:**
+- `label: "cited_fact" | "inference"` added to every cited key_point and risk in the LLM JSON schema
+- System prompt updated for all four providers (OpenAI, Anthropic, Gemini, Ollama)
+- UI: neutral "Inferred" badge shown next to inference items in `AIBriefCard`; cited_fact items render cleanly without a badge
+- `backfill_brief_labels` Celery task: classifies existing cited points in-place — one compact LLM call per brief (all points batched), updates JSONB with `flag_modified`, no brief re-generation
+- `POST /api/admin/backfill-labels` endpoint + "Backfill Fact/Inference Labels" button in Admin panel
+- `unlabeled_briefs` counter added to `/api/admin/stats` and pipeline breakdown table
+
+**Admin Panel Cleanup:**
+- Manual Controls split into two sections: always-visible recurring controls (Poll, Members, Trends, Actions, Resume) and a collapsible **Maintenance** section for one-time backfill tasks
+- Maintenance section header shows "⚠ action needed" when any backfill has a non-zero count
+
 ### v0.2.2 — Sponsor Linking & Search Fixes
 - **Root cause fixed:** Congress.gov list API does not return sponsor data — only the detail endpoint does. Poller now calls the detail endpoint for each new bill to get the sponsor and populate `bill.sponsor_id`
 - **Backfill task:** `backfill_sponsor_ids` Celery task + `/api/admin/backfill-sponsors` endpoint + "Backfill Sponsors" button in Admin UI — fixes existing bills with `NULL` sponsor_id (~10 req/sec, ~3 min for 1,600 bills)

PocketVeto — Feature Roadmap.md

@@ -31,6 +31,10 @@
 - [x] Backfill All Actions — admin task to fetch action history for all pre-existing bills
 - [x] Notifications (Phase 1) — ntfy dispatch, RSS feed, per-user settings UI, 5-min dispatcher beat task
 - [x] Brief Regeneration UI — admin button to delete existing briefs for a bill and re-queue LLM processing. Useful for improving citation/diff logic without a full re-poll. (Backend reprocess endpoint already exists.)
+- [x] Follow Modes — `neutral | pocket_veto | pocket_boost` on the `follows` table; FollowButton mode selector with descriptions and tooltips
+- [x] Public Browsing — unauthenticated guests browse bills, members, topics, and trending dashboard; AuthModal gates follow/interactive actions; sidebar and nav adapt to guest state
+- [x] Draft Letter Generator — collapsible panel on bill detail pages; select up to 3 cited brief points, stance auto-fills from follow mode, recipient derived from chamber, ZIP optional and never stored; calls configured LLM provider
+- [x] Bill Text Status Indicators — BillCard shows Brief / Pending / No text badge; backed by a single batch query on the list endpoint
 
 ---
 
@@ -40,8 +44,8 @@
 
 ### Phase 2 — High Impact *(can run in parallel after Phase 1)*
 
-- [ ] **Change-driven Alerts** — emit `notification_event` from poller/document fetcher on material changes: new doc version, substitute text, committee report, vote scheduled/result. Filter out procedural-only action text. Fan out to ntfy + RSS.
+- [x] **Change-driven Alerts** — milestone keywords expanded (markup, conference, referral tier); topic followers now receive bill_updated milestone events; committee referral events delivered to pocket_veto/boost but suppressed for neutral; all three follow types covered for both tiers.
-- [ ] **Fact vs Inference Labeling** — add `label: "cited_fact" | "inference"` + optional `confidence` field to each `key_point` and `risk` in the LLM JSON schema. Prompt engineering change + BillBrief schema migration. UI: small badge on each bullet (no color politics — neutral labels only).
+- [x] **Fact vs Inference Labeling** — `label: "cited_fact" | "inference"` on every cited key_point and risk; prompt engineering updated for all providers; "Inferred" badge in citation UI; backfill task for existing briefs.
 
 ---
 
@@ -76,10 +80,9 @@
 - [ ] **Source Viewer Option B** — in-app bill text viewer with cited passage highlighted and scroll-to-anchor. Deferred pending UX review of Option A (GovInfo link).
 - [ ] **Raw Diff Panel** — Python `difflib` diff between stored document versions, shown as collapsible "Raw Changes" below amendment brief. Zero API calls. Deferred — AI amendment brief is the primary "what changed" story.
 - [ ] **Shareable Collection Subscriptions** — "Follow this collection" mechanic so other users can subscribe to a public collection and get its bills added to their feed.
-- [ ] Pocket Veto mode (follow stance) — toggle on a bill to treat it as “I don’t want this to pass”; adds to watchlist and triggers milestone alerts (committee report-out, calendared, vote scheduled, passed chamber, etc.)
+- [x] Pocket Veto mode (follow stance) — toggle on a bill to treat it as “I don’t want this to pass”; adds to watchlist and triggers milestone alerts
-- [ ] Pocket Veto notification rules — alert only on advancement milestones + failure outcomes (failed committee / failed floor / stalled)
+- [x] Follow modes — Neutral + Pocket Veto + Pocket Boost; FollowButton is a mode selector with explanation tooltips
-- [ ] Follow modes — support Neutral (normal follow) + Pocket Veto now; optional Pocket Boost later
+- [ ] Pocket Veto notification rules — alert only on advancement milestones + failure outcomes (failed committee / failed floor / stalled) — notification dispatcher needs to filter by follow_mode
-- [ ] UI: FollowButton becomes FollowMode selector (Neutral / Pocket Veto) with explanation tooltip
 
 
 ### PocketVeto function
@@ -162,3 +165,51 @@ Then alert rules can be:
 - pocket_boost: “action points” + milestones
     
 
+Yes — that’s a solid idea **if it’s done as a “welcome + how it works” nudge**, not an annoying pop-up that blocks the UI.
+
+A toast can work, but for a first-time user you’ll usually get better results with a **dismissible banner** or a **one-time “welcome” card** on the dashboard, because:
+
+* toasts disappear (people miss them)
+* first-run onboarding usually needs at least one click (“Got it” / “Start here”)
+
+### Best pattern (low effort, high impact)
+
+**First visit → show a dismissible Welcome card/banner** (top of dashboard) with:
+
+* 1 sentence purpose
+* 3 bullets of key features
+* 2 buttons: **“Add my first follow”** and **“See a demo bill”**
+* “Don’t show again” checkbox (or implicit on dismiss)
+
+You can still use a toast, but make it:
+
+* sticky until dismissed
+* or paired with a banner/card
+
+### What it should say (copy you can paste)
+
+**Title:** Welcome to PocketVeto
+**Body (tight):**
+
+* Follow bills, members, or topics (low-noise)
+* See *what changed* in plain English
+* Verify every claim with **Back to Source** citations
+  **Buttons:** “Add a follow” | “Load demo”
+
+### Implementation detail (no creepy tracking)
+
+Store a simple flag:
+
+* `localStorage.setItem("pv_seen_welcome", "1")`
+
+Don’t store it server-side unless you already have user accounts and it’s part of preferences.
+
+### Backlog item (checkboxes)
+
+* [ ] First-visit welcome UI (banner/card + optional toast)
+* [ ] Dismiss + “don’t show again” (localStorage)
+* [ ] CTA: Add first follow
+* [ ] CTA: Load demo data (optional)
+* [ ] Link: “How it works” page/modal (optional)
+
+If you tell me your UI stack (sounds like Next.js + shadcn/ui), I can give you a drop-in component for a clean welcome card + toast behavior.

backend/app/api/admin.py

@@ -134,6 +134,23 @@ async def get_stats(
     bills_missing_actions = (await db.execute(
         text("SELECT COUNT(*) FROM bills WHERE actions_fetched_at IS NULL")
     )).scalar()
+    # Cited brief points (objects) that have no label yet
+    unlabeled_briefs = (await db.execute(
+        text("""
+            SELECT COUNT(*) FROM bill_briefs
+            WHERE (
+              key_points IS NOT NULL AND EXISTS (
+                SELECT 1 FROM jsonb_array_elements(key_points) AS p
+                WHERE jsonb_typeof(p) = 'object' AND (p->>'label') IS NULL
+              )
+            ) OR (
+              risks IS NOT NULL AND EXISTS (
+                SELECT 1 FROM jsonb_array_elements(risks) AS r
+                WHERE jsonb_typeof(r) = 'object' AND (r->>'label') IS NULL
+              )
+            )
+        """)
+    )).scalar()
     return {
         "total_bills": total_bills,
         "docs_fetched": docs_fetched,
@@ -146,6 +163,7 @@ async def get_stats(
         "bills_missing_sponsor": bills_missing_sponsor,
         "bills_missing_metadata": bills_missing_metadata,
         "bills_missing_actions": bills_missing_actions,
+        "unlabeled_briefs": unlabeled_briefs,
         "remaining": total_bills - total_briefs,
     }
 
@@ -204,6 +222,14 @@ async def backfill_metadata(current_user: User = Depends(get_current_admin)):
     return {"task_id": task.id, "status": "queued"}
 
 
+@router.post("/backfill-labels")
+async def backfill_labels(current_user: User = Depends(get_current_admin)):
+    """Classify existing cited brief points as fact or inference without re-generating briefs."""
+    from app.workers.llm_processor import backfill_brief_labels
+    task = backfill_brief_labels.delay()
+    return {"task_id": task.id, "status": "queued"}
+
+
 @router.post("/resume-analysis")
 async def resume_analysis(current_user: User = Depends(get_current_admin)):
     """Re-queue LLM processing for docs with no brief, and document fetching for bills with no doc."""

backend/app/services/llm_service.py

@@ -22,10 +22,10 @@ Always respond with valid JSON matching exactly this schema:
 {
   "summary": "2-4 paragraph plain-language summary of what this bill does",
   "key_points": [
-    {"text": "specific concrete fact", "citation": "Section X(y)", "quote": "verbatim excerpt from bill ≤80 words"}
+    {"text": "specific concrete fact", "citation": "Section X(y)", "quote": "verbatim excerpt from bill ≤80 words", "label": "cited_fact"}
   ],
   "risks": [
-    {"text": "legitimate concern or challenge", "citation": "Section X(y)", "quote": "verbatim excerpt from bill ≤80 words"}
+    {"text": "legitimate concern or challenge", "citation": "Section X(y)", "quote": "verbatim excerpt from bill ≤80 words", "label": "cited_fact"}
   ],
   "deadlines": [{"date": "YYYY-MM-DD or null", "description": "what happens on this date"}],
   "topic_tags": ["healthcare", "taxation"]
@@ -35,10 +35,14 @@ Rules:
 - summary: Explain WHAT the bill does, not whether it is good or bad. Be factual and complete.
 - key_points: 5-10 specific, concrete things the bill changes, authorizes, or appropriates. \
 Each item MUST include "text" (your claim), "citation" (the section number, e.g. "Section 301(a)(2)"), \
-and "quote" (a verbatim excerpt of ≤80 words from that section that supports your claim).
+"quote" (a verbatim excerpt of ≤80 words from that section that supports your claim), and "label".
 - risks: Legitimate concerns from any perspective — costs, implementation challenges, \
 constitutional questions, unintended consequences. Include at least 2 even for benign bills. \
-Each item MUST include "text", "citation", and "quote" just like key_points.
+Each item MUST include "text", "citation", "quote", and "label" just like key_points.
+- label: "cited_fact" if the claim is directly and explicitly stated in the quoted text. \
+"inference" if the claim is an analytical interpretation, projection, or implication that goes \
+beyond what the text literally says (e.g. projected costs, likely downstream effects, \
+constitutional questions). When in doubt, use "inference".
 - deadlines: Only include if explicitly stated in the text. Use null for date if a deadline \
 is mentioned without a specific date. Empty list if none.
 - topic_tags: 3-8 lowercase tags. Prefer these standard tags: healthcare, taxation, defense, \
@@ -90,10 +94,10 @@ Always respond with valid JSON matching exactly this schema:
 {
   "summary": "2-3 paragraph plain-language description of what changed in this version",
   "key_points": [
-    {"text": "specific change", "citation": "Section X(y)", "quote": "verbatim excerpt from new version ≤80 words"}
+    {"text": "specific change", "citation": "Section X(y)", "quote": "verbatim excerpt from new version ≤80 words", "label": "cited_fact"}
   ],
   "risks": [
-    {"text": "new concern introduced by this change", "citation": "Section X(y)", "quote": "verbatim excerpt from new version ≤80 words"}
+    {"text": "new concern introduced by this change", "citation": "Section X(y)", "quote": "verbatim excerpt from new version ≤80 words", "label": "cited_fact"}
   ],
   "deadlines": [{"date": "YYYY-MM-DD or null", "description": "new deadline added"}],
   "topic_tags": ["healthcare", "taxation"]
@@ -103,9 +107,12 @@ Rules:
 - summary: Focus ONLY on what is different from the previous version. Be specific.
 - key_points: List concrete additions, removals, or modifications in this version. \
 Each item MUST include "text" (your claim), "citation" (the section number, e.g. "Section 301(a)(2)"), \
-and "quote" (a verbatim excerpt of ≤80 words from the NEW version that supports your claim).
+"quote" (a verbatim excerpt of ≤80 words from the NEW version that supports your claim), and "label".
 - risks: Only include risks that are new or changed relative to the previous version. \
-Each item MUST include "text", "citation", and "quote" just like key_points.
+Each item MUST include "text", "citation", "quote", and "label" just like key_points.
+- label: "cited_fact" if the claim is directly and explicitly stated in the quoted text. \
+"inference" if the claim is an analytical interpretation, projection, or implication that goes \
+beyond what the text literally says. When in doubt, use "inference".
 - deadlines: Only new or changed deadlines. Empty list if none.
 - topic_tags: Same standard tags as before — include any new topics this version adds.
 

backend/app/workers/congress_poller.py

@@ -338,12 +338,29 @@ def _update_bill_if_changed(db, existing: Bill, parsed: dict) -> bool:
         from app.workers.notification_utils import (
             emit_bill_notification,
             emit_member_follow_notifications,
+            emit_topic_follow_notifications,
             is_milestone_action,
+            is_referral_action,
+        )
+        action_text = parsed.get("latest_action_text", "")
+        is_milestone = is_milestone_action(action_text)
+        is_referral = not is_milestone and is_referral_action(action_text)
+        if is_milestone or is_referral:
+            tier = "progress" if is_milestone else "referral"
+            emit_bill_notification(db, existing, "bill_updated", action_text, milestone_tier=tier)
+            emit_member_follow_notifications(db, existing, "bill_updated", action_text, milestone_tier=tier)
+            # Topic followers — pull tags from the bill's latest brief
+            from app.models.brief import BillBrief
+            latest_brief = (
+                db.query(BillBrief)
+                .filter_by(bill_id=existing.bill_id)
+                .order_by(BillBrief.created_at.desc())
+                .first()
+            )
+            topic_tags = latest_brief.topic_tags or [] if latest_brief else []
+            emit_topic_follow_notifications(
+                db, existing, "bill_updated", action_text, topic_tags, milestone_tier=tier
             )
-        if is_milestone_action(parsed.get("latest_action_text", "")):
-            action_text = parsed["latest_action_text"]
-            emit_bill_notification(db, existing, "bill_updated", action_text)
-            emit_member_follow_notifications(db, existing, "bill_updated", action_text)
     return changed
 
 

backend/app/workers/llm_processor.py

@@ -181,6 +181,113 @@ def backfill_brief_citations(self):
         db.close()
 
 
+@celery_app.task(bind=True, name="app.workers.llm_processor.backfill_brief_labels")
+def backfill_brief_labels(self):
+    """
+    Add fact/inference labels to existing cited brief points without re-generating them.
+    Sends one compact classification call per brief (all unlabeled points batched).
+    Skips briefs already fully labeled and plain-string points (no quote to classify).
+    """
+    import json
+    from sqlalchemy.orm.attributes import flag_modified
+    from app.models.setting import AppSetting
+
+    db = get_sync_db()
+    try:
+        unlabeled_ids = db.execute(text("""
+            SELECT id FROM bill_briefs
+            WHERE (
+              key_points IS NOT NULL AND EXISTS (
+                SELECT 1 FROM jsonb_array_elements(key_points) AS p
+                WHERE jsonb_typeof(p) = 'object' AND (p->>'label') IS NULL
+              )
+            ) OR (
+              risks IS NOT NULL AND EXISTS (
+                SELECT 1 FROM jsonb_array_elements(risks) AS r
+                WHERE jsonb_typeof(r) = 'object' AND (r->>'label') IS NULL
+              )
+            )
+        """)).fetchall()
+
+        total = len(unlabeled_ids)
+        updated = 0
+        skipped = 0
+
+        prov_row = db.get(AppSetting, "llm_provider")
+        model_row = db.get(AppSetting, "llm_model")
+        provider = get_llm_provider(
+            prov_row.value if prov_row else None,
+            model_row.value if model_row else None,
+        )
+
+        for row in unlabeled_ids:
+            brief = db.get(BillBrief, row.id)
+            if not brief:
+                skipped += 1
+                continue
+
+            # Collect all unlabeled cited points across both fields
+            to_classify: list[tuple[str, int, dict]] = []
+            for field_name in ("key_points", "risks"):
+                for i, p in enumerate(getattr(brief, field_name) or []):
+                    if isinstance(p, dict) and p.get("label") is None:
+                        to_classify.append((field_name, i, p))
+
+            if not to_classify:
+                skipped += 1
+                continue
+
+            lines = [
+                f'{i + 1}. TEXT: "{p["text"]}" | QUOTE: "{p.get("quote", "")}"'
+                for i, (_, __, p) in enumerate(to_classify)
+            ]
+            prompt = (
+                "Classify each item as 'cited_fact' or 'inference'.\n"
+                "cited_fact = the claim is explicitly and directly stated in the quoted text.\n"
+                "inference = analytical interpretation, projection, or implication not literally stated.\n\n"
+                "Return ONLY a JSON array of strings, one per item, in order. No explanation.\n\n"
+                "Items:\n" + "\n".join(lines)
+            )
+
+            try:
+                raw = provider.generate_text(prompt).strip()
+                if raw.startswith("```"):
+                    raw = raw.split("```")[1]
+                    if raw.startswith("json"):
+                        raw = raw[4:]
+                labels = json.loads(raw.strip())
+                if not isinstance(labels, list) or len(labels) != len(to_classify):
+                    logger.warning(f"Brief {brief.id}: label count mismatch, skipping")
+                    skipped += 1
+                    continue
+            except Exception as exc:
+                logger.warning(f"Brief {brief.id}: classification failed: {exc}")
+                skipped += 1
+                time.sleep(0.5)
+                continue
+
+            fields_modified: set[str] = set()
+            for (field_name, point_idx, _), label in zip(to_classify, labels):
+                if label in ("cited_fact", "inference"):
+                    getattr(brief, field_name)[point_idx]["label"] = label
+                    fields_modified.add(field_name)
+
+            for field_name in fields_modified:
+                flag_modified(brief, field_name)
+
+            db.commit()
+            updated += 1
+            time.sleep(0.2)
+
+        logger.info(
+            f"backfill_brief_labels: {total} briefs found, "
+            f"{updated} updated, {skipped} skipped"
+        )
+        return {"total": total, "updated": updated, "skipped": skipped}
+    finally:
+        db.close()
+
+
 @celery_app.task(bind=True, name="app.workers.llm_processor.resume_pending_analysis")
 def resume_pending_analysis(self):
     """

backend/app/workers/notification_dispatcher.py

@@ -94,6 +94,13 @@ def dispatch_notifications(self):
                 db.commit()
                 continue
 
+            # Referral-tier events (committee referrals) are noisy for neutral follows;
+            # pocket_veto and pocket_boost users want them as early warnings
+            if follow_mode == "neutral" and (event.payload or {}).get("milestone_tier") == "referral":
+                event.dispatched_at = now
+                db.commit()
+                continue
+
             prefs = user.notification_prefs or {}
             ntfy_url = prefs.get("ntfy_topic_url", "").strip()
             ntfy_auth_method = prefs.get("ntfy_auth_method", "none")

backend/app/workers/notification_utils.py

@@ -12,6 +12,13 @@ _MILESTONE_KEYWORDS = [
     "reported by", "discharged",
     "placed on",        # placed on calendar
     "cloture", "roll call",
+    "markup",           # markup session scheduled/completed
+    "conference",       # conference committee activity
+]
+
+# Committee referral — meaningful for pocket_veto/boost but noisy for neutral
+_REFERRAL_KEYWORDS = [
+    "referred to",
 ]
 
 # Events created within this window for the same (user, bill, event_type) are suppressed
@@ -23,7 +30,12 @@ def is_milestone_action(action_text: str) -> bool:
     return any(kw in t for kw in _MILESTONE_KEYWORDS)
 
 
-def _build_payload(bill, action_summary: str) -> dict:
+def is_referral_action(action_text: str) -> bool:
+    t = (action_text or "").lower()
+    return any(kw in t for kw in _REFERRAL_KEYWORDS)
+
+
+def _build_payload(bill, action_summary: str, milestone_tier: str = "progress") -> dict:
     from app.config import settings
     base_url = (settings.PUBLIC_URL or settings.LOCAL_URL).rstrip("/")
     return {
@@ -31,6 +43,7 @@ def _build_payload(bill, action_summary: str) -> dict:
         "bill_label": f"{bill.bill_type.upper()} {bill.bill_number}",
         "brief_summary": (action_summary or "")[:300],
         "bill_url": f"{base_url}/bills/{bill.bill_id}",
+        "milestone_tier": milestone_tier,
     }
 
 
@@ -45,7 +58,9 @@ def _is_duplicate(db, user_id: int, bill_id: str, event_type: str) -> bool:
     ).filter(NotificationEvent.created_at > cutoff).first() is not None
 
 
-def emit_bill_notification(db, bill, event_type: str, action_summary: str) -> int:
+def emit_bill_notification(
+    db, bill, event_type: str, action_summary: str, milestone_tier: str = "progress"
+) -> int:
     """Create NotificationEvent rows for every user following this bill. Returns count."""
     from app.models.follow import Follow
     from app.models.notification import NotificationEvent
@@ -54,7 +69,7 @@ def emit_bill_notification(db, bill, event_type: str, action_summary: str) -> in
     if not followers:
         return 0
 
-    payload = _build_payload(bill, action_summary)
+    payload = _build_payload(bill, action_summary, milestone_tier)
     count = 0
     for follow in followers:
         if _is_duplicate(db, follow.user_id, bill.bill_id, event_type):
@@ -71,7 +86,9 @@ def emit_bill_notification(db, bill, event_type: str, action_summary: str) -> in
     return count
 
 
-def emit_member_follow_notifications(db, bill, event_type: str, action_summary: str) -> int:
+def emit_member_follow_notifications(
+    db, bill, event_type: str, action_summary: str, milestone_tier: str = "progress"
+) -> int:
     """Notify users following the bill's sponsor (dedup prevents double-alerts for bill+member followers)."""
     if not bill.sponsor_id:
         return 0
@@ -83,7 +100,7 @@ def emit_member_follow_notifications(db, bill, event_type: str, action_summary:
     if not followers:
         return 0
 
-    payload = _build_payload(bill, action_summary)
+    payload = _build_payload(bill, action_summary, milestone_tier)
     count = 0
     for follow in followers:
         if _is_duplicate(db, follow.user_id, bill.bill_id, event_type):
@@ -100,7 +117,10 @@ def emit_member_follow_notifications(db, bill, event_type: str, action_summary:
     return count
 
 
-def emit_topic_follow_notifications(db, bill, event_type: str, action_summary: str, topic_tags: list) -> int:
+def emit_topic_follow_notifications(
+    db, bill, event_type: str, action_summary: str, topic_tags: list,
+    milestone_tier: str = "progress",
+) -> int:
     """Notify users following any of the bill's topic tags."""
     if not topic_tags:
         return 0
@@ -120,7 +140,7 @@ def emit_topic_follow_notifications(db, bill, event_type: str, action_summary: s
     if not followers:
         return 0
 
-    payload = _build_payload(bill, action_summary)
+    payload = _build_payload(bill, action_summary, milestone_tier)
     count = 0
     for follow in followers:
         if _is_duplicate(db, follow.user_id, bill.bill_id, event_type):

frontend/app/settings/page.tsx

@@ -17,6 +17,9 @@ import {
   Bell,
   Shield,
   Zap,
+  ChevronDown,
+  ChevronRight,
+  Wrench,
 } from "lucide-react";
 import Link from "next/link";
 import { settingsAPI, adminAPI, notificationsAPI, type AdminUser, type LLMModel, type ApiHealthResult } from "@/lib/api";
@@ -134,6 +137,7 @@ export default function SettingsPage() {
   const [taskIds, setTaskIds] = useState<Record<string, string>>({});
   const [taskStatuses, setTaskStatuses] = useState<Record<string, "running" | "done" | "error">>({});
   const [confirmDelete, setConfirmDelete] = useState<number | null>(null);
+  const [showMaintenance, setShowMaintenance] = useState(false);
 
   const testLLM = async () => {
     setTesting(true);
@@ -316,6 +320,7 @@ export default function SettingsPage() {
                 { label: "AI briefs generated", value: stats.briefs_generated, color: "text-green-600 dark:text-green-400", icon: "✅" },
                 { label: "Pending LLM analysis", value: stats.pending_llm, color: stats.pending_llm > 0 ? "text-amber-600 dark:text-amber-400" : "text-muted-foreground", icon: "🔄", action: stats.pending_llm > 0 ? "Resume Analysis" : undefined },
                 { label: "Briefs missing citations", value: stats.uncited_briefs, color: stats.uncited_briefs > 0 ? "text-amber-600 dark:text-amber-400" : "text-muted-foreground", icon: "⚠️", action: stats.uncited_briefs > 0 ? "Backfill Citations" : undefined },
+                { label: "Briefs with unlabeled points", value: stats.unlabeled_briefs, color: stats.unlabeled_briefs > 0 ? "text-amber-600 dark:text-amber-400" : "text-muted-foreground", icon: "🏷️", action: stats.unlabeled_briefs > 0 ? "Backfill Labels" : undefined },
               ].map(({ label, value, color, icon, note, action }) => (
                 <div key={label} className="flex items-center justify-between py-2.5 gap-3">
                   <div className="flex items-center gap-2 min-w-0">
@@ -637,82 +642,9 @@ export default function SettingsPage() {
       {/* Manual Controls */}
       <section className="bg-card border border-border rounded-lg p-6 space-y-4">
         <h2 className="font-semibold">Manual Controls</h2>
-        <div className="divide-y divide-border">
+
-          {([
+        {(() => {
-            {
+          type ControlItem = {
-              key: "poll",
-              name: "Trigger Poll",
-              description: "Check Congress.gov for newly introduced or updated bills. Runs automatically on a schedule — use this to force an immediate sync.",
-              fn: adminAPI.triggerPoll,
-              status: "on-demand",
-            },
-            {
-              key: "members",
-              name: "Sync Members",
-              description: "Refresh all member profiles from Congress.gov including biography, current term, leadership roles, and contact information.",
-              fn: adminAPI.triggerMemberSync,
-              status: "on-demand",
-            },
-            {
-              key: "trends",
-              name: "Calculate Trends",
-              description: "Score bill and member newsworthiness by counting recent news headlines and Google search interest. Updates the trend charts.",
-              fn: adminAPI.triggerTrendScores,
-              status: "on-demand",
-            },
-            {
-              key: "actions",
-              name: "Fetch Bill Actions",
-              description: "Download the full legislative history (votes, referrals, amendments) for recently active bills and populate the timeline view.",
-              fn: adminAPI.triggerFetchActions,
-              status: "on-demand",
-            },
-            {
-              key: "backfill-actions",
-              name: "Backfill All Action Histories",
-              description: "One-time catch-up: fetch action histories for all bills that were imported before this feature existed. Run once to populate timelines across your full bill archive.",
-              fn: adminAPI.backfillAllActions,
-              status: stats ? (stats.bills_missing_actions > 0 ? "needed" : "ok") : "on-demand",
-              count: stats?.bills_missing_actions,
-              countLabel: "bills missing action history",
-            },
-            {
-              key: "sponsors",
-              name: "Backfill Sponsors",
-              description: "Link bill sponsors that weren't captured during the initial import. Safe to re-run — skips bills that already have a sponsor.",
-              fn: adminAPI.backfillSponsors,
-              status: stats ? (stats.bills_missing_sponsor > 0 ? "needed" : "ok") : "on-demand",
-              count: stats?.bills_missing_sponsor,
-              countLabel: "bills missing sponsor",
-            },
-            {
-              key: "metadata",
-              name: "Backfill Dates & Links",
-              description: "Fill in missing introduced dates, chamber assignments, and congress.gov links by re-fetching bill detail from Congress.gov.",
-              fn: adminAPI.backfillMetadata,
-              status: stats ? (stats.bills_missing_metadata > 0 ? "needed" : "ok") : "on-demand",
-              count: stats?.bills_missing_metadata,
-              countLabel: "bills missing metadata",
-            },
-            {
-              key: "citations",
-              name: "Backfill Citations",
-              description: "Regenerate AI briefs that were created before inline source citations were added. Deletes the old brief and re-runs LLM analysis using the already-stored bill text — no new Congress.gov calls.",
-              fn: adminAPI.backfillCitations,
-              status: stats ? (stats.uncited_briefs > 0 ? "needed" : "ok") : "on-demand",
-              count: stats?.uncited_briefs,
-              countLabel: "briefs need regeneration",
-            },
-            {
-              key: "resume",
-              name: "Resume Analysis",
-              description: "Restart AI brief generation for bills where processing stalled or failed (e.g. after an LLM quota outage). Also re-queues document fetching for bills that have no text yet.",
-              fn: adminAPI.resumeAnalysis,
-              status: stats ? (stats.pending_llm > 0 ? "needed" : "ok") : "on-demand",
-              count: stats?.pending_llm,
-              countLabel: "bills pending analysis",
-            },
-          ] as Array<{
             key: string;
             name: string;
             description: string;
@@ -720,7 +652,9 @@ export default function SettingsPage() {
             status: "ok" | "needed" | "on-demand";
             count?: number;
             countLabel?: string;
-          }>).map(({ key, name, description, fn, status, count, countLabel }) => (
+          };
+
+          const renderRow = ({ key, name, description, fn, status, count, countLabel }: ControlItem) => (
             <div key={key} className="flex items-start gap-3 py-3.5">
               <div className={`w-2.5 h-2.5 rounded-full mt-1 shrink-0 ${
                 status === "ok" ? "bg-green-500"
@@ -757,13 +691,133 @@ export default function SettingsPage() {
                 disabled={taskStatuses[key] === "running"}
                 className="shrink-0 flex items-center gap-1.5 px-3 py-1.5 text-xs bg-muted hover:bg-accent rounded-md transition-colors font-medium disabled:opacity-50 disabled:cursor-not-allowed"
               >
-                {taskStatuses[key] === "running" ? (
+                {taskStatuses[key] === "running" ? <RefreshCw className="w-3 h-3 animate-spin" /> : "Run"}
-                  <RefreshCw className="w-3 h-3 animate-spin" />
-                ) : "Run"}
               </button>
             </div>
-          ))}
+          );
+
+          const recurring: ControlItem[] = [
+            {
+              key: "poll",
+              name: "Trigger Poll",
+              description: "Check Congress.gov for newly introduced or updated bills. Runs automatically on a schedule — use this to force an immediate sync.",
+              fn: adminAPI.triggerPoll,
+              status: "on-demand",
+            },
+            {
+              key: "members",
+              name: "Sync Members",
+              description: "Refresh all member profiles from Congress.gov including biography, current term, leadership roles, and contact information.",
+              fn: adminAPI.triggerMemberSync,
+              status: "on-demand",
+            },
+            {
+              key: "trends",
+              name: "Calculate Trends",
+              description: "Score bill and member newsworthiness by counting recent news headlines and Google search interest. Updates the trend charts.",
+              fn: adminAPI.triggerTrendScores,
+              status: "on-demand",
+            },
+            {
+              key: "actions",
+              name: "Fetch Bill Actions",
+              description: "Download the full legislative history (votes, referrals, amendments) for recently active bills and populate the timeline view.",
+              fn: adminAPI.triggerFetchActions,
+              status: "on-demand",
+            },
+            {
+              key: "resume",
+              name: "Resume Analysis",
+              description: "Restart AI brief generation for bills where processing stalled or failed (e.g. after an LLM quota outage). Also re-queues document fetching for bills that have no text yet.",
+              fn: adminAPI.resumeAnalysis,
+              status: stats ? (stats.pending_llm > 0 ? "needed" : "on-demand") : "on-demand",
+              count: stats?.pending_llm,
+              countLabel: "bills pending analysis",
+            },
+          ];
+
+          const maintenance: ControlItem[] = [
+            {
+              key: "backfill-actions",
+              name: "Backfill All Action Histories",
+              description: "One-time catch-up: fetch action histories for all bills that were imported before this feature existed.",
+              fn: adminAPI.backfillAllActions,
+              status: stats ? (stats.bills_missing_actions > 0 ? "needed" : "ok") : "on-demand",
+              count: stats?.bills_missing_actions,
+              countLabel: "bills missing action history",
+            },
+            {
+              key: "sponsors",
+              name: "Backfill Sponsors",
+              description: "Link bill sponsors that weren't captured during the initial import. Safe to re-run — skips bills that already have a sponsor.",
+              fn: adminAPI.backfillSponsors,
+              status: stats ? (stats.bills_missing_sponsor > 0 ? "needed" : "ok") : "on-demand",
+              count: stats?.bills_missing_sponsor,
+              countLabel: "bills missing sponsor",
+            },
+            {
+              key: "metadata",
+              name: "Backfill Dates & Links",
+              description: "Fill in missing introduced dates, chamber assignments, and congress.gov links by re-fetching bill detail from Congress.gov.",
+              fn: adminAPI.backfillMetadata,
+              status: stats ? (stats.bills_missing_metadata > 0 ? "needed" : "ok") : "on-demand",
+              count: stats?.bills_missing_metadata,
+              countLabel: "bills missing metadata",
+            },
+            {
+              key: "citations",
+              name: "Backfill Citations",
+              description: "Regenerate AI briefs created before inline source citations were added. Deletes the old brief and re-runs LLM analysis using already-stored bill text.",
+              fn: adminAPI.backfillCitations,
+              status: stats ? (stats.uncited_briefs > 0 ? "needed" : "ok") : "on-demand",
+              count: stats?.uncited_briefs,
+              countLabel: "briefs need regeneration",
+            },
+            {
+              key: "labels",
+              name: "Backfill Fact/Inference Labels",
+              description: "Classify existing cited brief points as fact or inference. One compact LLM call per brief — no re-generation of summaries or citations.",
+              fn: adminAPI.backfillLabels,
+              status: stats ? (stats.unlabeled_briefs > 0 ? "needed" : "ok") : "on-demand",
+              count: stats?.unlabeled_briefs,
+              countLabel: "briefs with unlabeled points",
+            },
+          ];
+
+          const maintenanceNeeded = maintenance.some((m) => m.status === "needed");
+
+          return (
+            <>
+              <div className="divide-y divide-border">
+                {recurring.map(renderRow)}
               </div>
+
+              {/* Maintenance subsection */}
+              <div className="border border-border rounded-md overflow-hidden">
+                <button
+                  onClick={() => setShowMaintenance((v) => !v)}
+                  className="w-full flex items-center justify-between px-4 py-3 text-sm font-medium bg-muted/50 hover:bg-muted transition-colors"
+                >
+                  <span className="flex items-center gap-2">
+                    <Wrench className="w-3.5 h-3.5 text-muted-foreground" />
+                    Maintenance
+                    {maintenanceNeeded && (
+                      <span className="text-xs font-normal text-red-600 dark:text-red-400">⚠ action needed</span>
+                    )}
+                  </span>
+                  {showMaintenance
+                    ? <ChevronDown className="w-4 h-4 text-muted-foreground" />
+                    : <ChevronRight className="w-4 h-4 text-muted-foreground" />}
+                </button>
+                {showMaintenance && (
+                  <div className="divide-y divide-border px-4">
+                    {maintenance.map(renderRow)}
+                  </div>
+                )}
+              </div>
+            </>
+          );
+        })()}
       </section>
     </div>
   );

frontend/components/bills/AIBriefCard.tsx

@@ -31,6 +31,14 @@ function CitedItem({ point, icon, govinfo_url, openKey, activeKey, setActiveKey
       <div className="flex items-start gap-2">
         <span className="mt-0.5 shrink-0">{icon}</span>
         <span className="flex-1">{cited ? point.text : point}</span>
+        {cited && point.label === "inference" && (
+          <span
+            title="This point is an analytical interpretation, not a literal statement from the bill text"
+            className="shrink-0 text-[10px] px-1.5 py-0.5 rounded border border-border text-muted-foreground font-sans leading-none"
+          >
+            Inferred
+          </span>
+        )}
         {cited && (
           <button
             onClick={() => setActiveKey(isOpen ? null : openKey)}

frontend/lib/api.ts

@@ -174,6 +174,7 @@ export interface AnalysisStats {
   bills_missing_sponsor: number;
   bills_missing_metadata: number;
   bills_missing_actions: number;
+  unlabeled_briefs: number;
   remaining: number;
 }
 
@@ -236,6 +237,8 @@ export const adminAPI = {
     apiClient.post("/api/admin/backfill-all-actions").then((r) => r.data),
   backfillMetadata: () =>
     apiClient.post("/api/admin/backfill-metadata").then((r) => r.data),
+  backfillLabels: () =>
+    apiClient.post("/api/admin/backfill-labels").then((r) => r.data),
   resumeAnalysis: () =>
     apiClient.post("/api/admin/resume-analysis").then((r) => r.data),
   getApiHealth: () =>

frontend/lib/types.ts

@@ -58,6 +58,7 @@ export interface CitedPoint {
   text: string;
   citation: string;
   quote: string;
+  label?: "cited_fact" | "inference";
 }
 
 export interface BriefSchema {