fix: resolve stash conflict and update roadmap to v2

Resolved merge conflict in exercise_card.html (kept suggested
reps/weight pre-population logic) and updated roadmap.md with
the new v2 improvement plan.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app/templates/partials/exercise_card.html

@@ -32,6 +32,13 @@
     <!-- Inline logging (Phase 4) -->
     {% if active_profile %}
     <div id="logs-exercise-{{ exercise.id }}">
+        {% if existing_logs and existing_logs[exercise.id] %}
+            {% set suggested_reps = existing_logs[exercise.id][-1].reps_completed %}
+            {% set suggested_weight = existing_logs[exercise.id][-1].weight_used %}
+        {% elif suggestions and suggestions[exercise.id] %}
+            {% set suggested_reps = suggestions[exercise.id].suggested_reps %}
+            {% set suggested_weight = suggestions[exercise.id].suggested_weight %}
+        {% endif %}
         {% if existing_logs and existing_logs[exercise.id] %}
             {% set logs = existing_logs[exercise.id] %}
             {% set exercise_id = exercise.id %}

docs/roadmap.md

@@ -1,30 +1,127 @@
 # SneakySwole Roadmap
 
-## Completed
+## V2 Improvements
 
-### Phase 1: Scaffold & Infrastructure
+Each improvement is implemented as a separate branch/PR. Implementation order matters — auth removal goes first since other features depend on the simplified profile flow.
-FastAPI project structure, Dockerfile + docker-compose.yaml, Pico CSS dark theme base template, `.env` config, structlog logging, health check endpoint.
 
-### Phase 2: Data Layer & Seeding
+---
-8-table SQLite schema via SQLModel + Alembic migrations. YAML-driven seed script (`config/exercises.yaml`, `config/user_programs.yaml`). Service layer for all DB access. Admin user auto-created from `.env` with bcrypt.
 
-### Phase 3: Workout UI
+### 1. Easier Profile Selection
-Admin login (bcrypt + signed session cookies), profile switcher, workout day viewer with warmups and exercise cards, HTMX-powered exercise browser with search/filter. NavContextMiddleware for automatic template context.
+**Branch:** `feature/remove-auth`
 
-### Phase 4: Logging & Tracking
+Strip all authentication entirely — this is a homelab app, no need for login. When the app opens, the landing page (`/`) shows a profile picker with cards for each user. Selecting a profile sets an `active_profile_id` cookie that persists across sessions (per-device, until the user switches profiles or clears cookies).
-Inline set logging from the workout day view via HTMX. Auto-created workout sessions. Log history with per-session detail view. Edit and delete log entries.
 
-### Phase 5: Progression & Analytics
+**What gets removed:**
-Auto-progression engine (+reps/+weight/deload rules), 4-week schedule calendar, progress dashboard with Chart.js charts, per-exercise progress pages with suggestions.
+- Admin login flow (login page, bcrypt password verification, session tokens)
+- `is_admin` and `password_hash` fields from the users table
+- `auth_service.py`, `auth.py` routes, `login.html` template
+- `bcrypt` and `itsdangerous` dependencies
+- Auth dependency (`get_current_admin_user`) from all route handlers
+
+**What gets added/changed:**
+- `/` becomes the profile picker page (cards with name, height, weight, and "Select" button)
+- `require_active_profile()` dependency replaces auth — just checks for the cookie
+- `NoProfileSelectedError` handler redirects to `/` if no profile is set
+- `NavContextMiddleware` simplified to only read the `active_profile_id` cookie
+- Nav bar always shows profile switcher dropdown (no admin gating)
+
+**Edge cases:**
+- No profiles exist → show "Create Profile" link
+- Cookie references a deleted profile → middleware returns None, redirect to picker
+- Existing DB with admin row → migration deletes it
+
+---
+
+### 2. Schedule Adjustment — "Workout Now" Flow
+**Branch:** `feature/workout-now` | **Depends on:** #1 merged
+
+Replace the `/workouts` page (static list of all 4 days) with a smart "Workout Now" page that auto-recommends the next workout in the Push → Pull → Lower → Full Body cycle based on the user's last completed session.
+
+**How it works:**
+1. User selects profile → navigates to `/workouts`
+2. App queries the most recent `workout_session` for that profile
+3. Determines next day in cycle (e.g., last did Push → recommend Pull). No history defaults to Push.
+4. Page shows all 4 workout types as cards, with the recommended one visually highlighted (`<mark>Recommended Next</mark>`, primary button vs outline for others)
+5. User clicks "Start Workout" on any card → goes to that workout's exercise page
+6. Once on a workout, a "Change Workout" button is visible to switch types (soft lock — not hard-locked, but the UI keeps you focused on the chosen workout)
+
+**Card-based approach (no JS):** The recommendation is shown via highlighted cards rather than a dropdown/select. This follows the project's HTMX-first principle — no JavaScript needed.
+
+**Display context:** Show "Last workout: Pull on Mar 10" message on the page so the user understands the recommendation.
+
+**Edge cases:**
+- Same-day multiple workout types → creates separate sessions (existing behavior, correct)
+- No profile selected → show "pick a profile first" message with link to `/`
+
+---
+
+### 3. Automatic Scaled Workout Reps / Sets / Weights — 6→12 Rep Ladder
+**Branch:** `feature/rep-ladder-progression` | **Depends on:** #1 merged
+
+Replace the current wk1/wk4 target system with a fixed, universal rep ladder progression. Every exercise follows the same pattern:
+
+**Progression model:**
+- **Rep ladder:** 6 → 8 → 10 → 12 reps at current weight
+- **Weight increase:** At 12 reps with all sets marked "felt easy" → +5 lbs, reset to 6 reps
+- **Hold:** If not all sets felt easy at current reps → stay at current reps/weight
+- **Climb:** If all sets felt easy but below 12 reps → move to next step in ladder
+- **Deload:** After 4+ consecutive sessions struggling (none felt easy) → reduce weight by 20%, reset to 6 reps
+- **Always 3 sets** per exercise
+
+**Schema change:** Simplify `user_exercise_programs` table — keep only `starting_weight` per user/exercise (migrated from existing `wk1_weight` data). Drop `wk1_reps`, `wk4_reps`, `wk4_weight` columns. The existing 40 rows of programming data (20 exercises × 2 users) are preserved as starting weights.
+
+**Starting weight behavior:**
+- Exercise with no log history → suggest starting_weight at 6 reps, 3 sets
+- Exercise with log history → progression engine picks up from last logged values
+
+**UI change:** Replace the Week 1/Week 4 target display on exercise cards with a rep ladder progress bar showing the 4 steps (6, 8, 10, 12) with the current position filled in.
+
+**Bodyweight exercises** (weight = "BW"): Weight increase is skipped. User climbs the rep ladder and holds at 12 reps.
+
+**Data continuity:** The 55 existing workout log entries are unchanged and feed directly into the new progression engine. The progression_service.py gets a full rewrite of `get_suggestion()` with the new ladder logic.
+
+---
+
+### 4. Export Workout History as CSV
+**Branch:** `feature/csv-export` | **Depends on:** #1 merged
+
+Add a CSV export section to the dashboard page with a date range picker (default: last 30 days) and a "Download CSV" button.
+
+**CSV format — one row per set:**
+| Column | Description |
+|--------|-------------|
+| date | Workout session date (ISO format) |
+| workout_type | Push / Pull / Lower / Full Body |
+| exercise | Exercise name |
+| set_number | Set number within that exercise |
+| reps | Reps completed |
+| weight | Weight used |
+| felt_easy | Yes / No |
+
+**Implementation:**
+- `GET /dashboard/export?start_date=YYYY-MM-DD&end_date=YYYY-MM-DD` returns a CSV file download
+- Uses standard HTML form GET (not HTMX — HTMX can't trigger file downloads)
+- `StreamingResponse` with `Content-Disposition: attachment` header
+- Filename: `sneakyswole_{profile_name}_{start}_to_{end}.csv`
+- Date range defaults: start = 30 days ago, end = today. Invalid dates fall back to defaults.
+
+**Edge cases:**
+- No data in range → CSV with headers only (valid empty CSV)
+- No profile selected → redirect to dashboard
+- Large date ranges are fine — even a year of heavy use is ~3,000 rows
+
+---
+
+## Implementation Order
+
+```
+1. Remove Auth ──┬──> 2. Workout Now
+                 ├──> 3. Rep Ladder Progression  (can be parallel with 4)
+                 └──> 4. CSV Export              (can be parallel with 3)
+```
 
 ---
 
 ## Future Ideas
-
-- Multi-user login (replace profile switcher with individual logins)
-- REST API for mobile clients
 - Exercise video/image attachments
 - Custom workout program builder
-- Export/import workout data (CSV/JSON)
-- Notifications/reminders
-- Social features (sharing workouts)