Add engine/history.go with GetTaskHistory and diff-style FormatTaskHistory
that compares consecutive change_log entries to show only what changed.
Add cmd/log.go command for full task history. Integrate last 5 history
entries into FormatTaskDetail (info view) as a "Recent Changes" section.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add annotations as JSON column on tasks table with Annotate/Denotate
methods and CLI commands. Add undo system backed by change_log with
lightweight undo_stack table (capped at 10 entries). All mutating CLI
commands (add, done, delete, modify, start, stop) now record undo
entries. Undo restores prior task state from change_log data.
Schema changes (in v1 migration):
- annotations TEXT column on tasks
- undo_stack table
- annotations field in change_log triggers
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds a persistent --dry-run flag that shows matched tasks without
performing mutations. Supported on done, delete, modify, start, and stop
commands. Also fixes preprocessArgs to skip flag-like args when
identifying commands, preventing flags from being treated as filters.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add relative date formatting (today, tomorrow, in 3d, etc.) for list and
detail views. Add structured feedback helpers for add/complete/delete
operations showing display IDs and parsed modifiers. Change Complete() to
return spawned recurring instance so callers can display recurrence info.
Add AppendTask to working set for immediate display ID assignment.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
opal version (and opal --version) prints version, commit, and build
date. Values are set via ldflags at build time; defaults to "dev" for
local builds. VERSION file added at 0.1.0.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
IMP-5: Replace strings.Contains(arg, ":") heuristic with an allowlist
of recognized attribute keys (ValidAttributeKeys). Colons in task
descriptions (URLs, "Meeting: topic") are no longer misinterpreted as
modifiers. Canonical key sets live in engine/keys.go and are shared
across parseAddArgs, ParseFilter, and ParseModifier. ParseModifier now
errors on unknown keys.
IMP-4: delete command now loads the working set and resolves display IDs
via GetTaskByDisplayID, matching the pattern used by done/modify.
IMP-6: All action commands (done, delete, modify, start, stop) now
return an error on no-match (stderr, exit 1). Previously done/delete
printed to stdout and exited 0; start/stop had no check at all.
Also adds requirements and design docs for the CLI UX improvements.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
import.meta.env.DEV is already true during `bun run dev` and false in
production builds, so a separate env var is unnecessary.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add --dev flag to `opal server start` that disables auth (injects
userID=1 for all requests) and exposes a /auth/dev-session endpoint,
so the frontend can develop against a real backend without OAuth
config. Remove VITE_MOCK_MODE and all mock data/branches from the
frontend stores. Add scripts/dev.sh to start both services locally.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Load config eagerly during server startup so sortByUrgency never
hits a nil config. Add nil-guard in BuildUrgencyCoefficients as
belt-and-suspenders defense. Fix OAuth callback to support both
GET and POST, and resolve issuer URLs properly with path.Dir.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
LoadConfig() tried to create directories and write opal.yml as a side
effect of loading config. On the server (where /etc/opal is in systemd
ReadOnlyPaths), this failed, returning nil. All internal GetConfig()
callers discarded the error, passing nil to BuildUrgencyCoefficients()
which panicked on nil dereference.
Redesign the config system with layered, read-only loading:
- Defaults (always present) → YAML file (if exists) → OPAL_ env vars
- LoadConfig never writes to the filesystem or returns nil
- File creation moved to explicit InitConfig() for CLI first-run/setup
- SaveConfig uses yaml.Marshal instead of manual field-by-field Viper
calls, eliminating the three-place maintenance burden
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove ThemeSwitcher from header (already accessible via settings)
- Increase pill padding and font size for better tap targets
- Guard non-cancelable touchmove preventDefault in SwipeAction
- Restyle settings page with grid-area layout and inline sign-out button
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Fix latent API bug where multi-word fields (RecurrenceDuration, ParentUUID,
CreatedAt) serialized as PascalCase, breaking the frontend. Add explicit
snake_case json tags and custom MarshalJSON/UnmarshalJSON on Task, Status,
and APIKey to emit unix timestamps and string status codes.
Add Urgency float64 as a derived field on Task, populated via
PopulateUrgency helper in all handlers before serialization. The report
engine's sortByUrgency now also retains the computed score.
Frontend updated with urgency type, color-coded badge in TaskItem, and
mock data values.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace flexbox layout with CSS grid using named grid-areas for responsive
content containment. Gutters collapse naturally on small screens via
min(--content-max-width, 100%). Anchor ReportPicker to its trigger button
using CSS anchor positioning instead of fixed viewport offsets.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extract CreateRecurringTask into engine package for reuse by both CLI
and API. Add POST /tasks/parse endpoint for CLI-style input parsing.
Remove FK constraint on change_log to preserve history after task
deletion. Update web frontend to filter completed tasks from view and
add mock mode support for development.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three holistic design directions with CSS custom properties, a theme
store persisted to localStorage, and a live switcher in both the header
(cycle button) and settings page (card selector). Also fixes checkbox
checkmark alignment and adds back navigation from settings.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When a task is completed via checkbox tap, it fades out and collapses
before being removed from the list, matching the swipe-to-complete
animation behavior.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace multi-page task management with single-screen layout: Header
with report picker at top, scrollable TaskList in the middle, and
InputBar with property pills fixed at the bottom. Owns state for
active report, task loading, input parsing, and task completion.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
TaskItem: remove onClick navigation, wrap in SwipeAction for
swipe-to-complete, update priority colors (H=red, M=amber, L=gray,
default=hidden), add due-today amber color.
TaskList: accept activeReport prop for context-aware empty states,
replace onToggle/onTaskClick with onComplete, make scrollable with
flex:1 and overflow-y:auto.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Implements right-swipe-to-complete with angle-based lock-in (horizontal
must exceed 2x vertical), 100px threshold, green checkmark background
reveal, and CSS transition for snap-back and completion animation.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
InputBar provides fixed-to-bottom text input with Enter to submit,
blur-delay (150ms) for pill interaction, and cursor-aware text
insertion. PropertyPills shows 8 modifier pills (Due, Pri, Project,
Tag, Recur, Scheduled, Wait, Until) on input focus.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Header shows active report name (left) with dropdown chevron and gear
icon linking to /settings (right). ReportPicker uses native Popover API
with 11 reports grouped into Common, Time-based, Recurring, and Archive.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Delete /tasks/new, /tasks/[uuid], /projects, /tags routes and BottomNav
component. Simplify layout to slot-only with 100dvh flexbox. Remove
nav-height CSS variable and .page padding rules.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add tasks.parse() and tasks.listByReport() to the API layer, and
loadReport() and parseAndCreate() to the tasks store with mock mode
support for the CLI-passthrough redesign.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace cobra.OnInitialize with PersistentPreRun pattern to fix initialization
issues with setup and server commands. This follows Cobra best practices and
allows subcommands to properly override initialization behavior.
Problem:
- OnInitialize runs globally before command parsing
- os.Args check for 'setup' was fragile and broke with flags
- Setup wizard failed on server: 'unable to open database file: no such file
or directory'
Solution:
- Use rootCmd.PersistentPreRun for initialization (inherited by all commands)
- setup and server commands override with their own PersistentPreRun
- Directory overrides still applied correctly in all cases
- Removes fragile os.Args parsing
Benefits:
- Works regardless of flag order
- Follows Cobra's intended design patterns
- Only 3 files modified (root.go, setup.go, server.go)
- Commands that need custom init (setup/server) simply override
- All other commands get automatic initialization
- Cleaner, more maintainable code
Testing:
- ✓ opal setup works without initialization errors
- ✓ opal list initializes database correctly
- ✓ First-run detection still works
- ✓ Directory overrides work with flags in any position
- ✓ Server command handles its own initialization
Display the default config and data directories before prompting
the user if they want to customize them. This gives better context
for the decision.
Before: Only showed directories if user chose not to customize
After: Shows directories first, then asks if user wants to change them
Implement a comprehensive setup wizard to improve onboarding and
configuration experience for both personal and server deployments.
Key features:
- Interactive wizard with profile selection (personal/server/custom)
- Quick setup mode with sensible defaults
- First-run detection with helpful welcome message
- Directory configuration with validation
- Server OAuth/JWT configuration with auto-generation
- Environment file creation for server deployments
- Template generators for systemd service and env files
New commands:
- opal setup # Interactive wizard
- opal setup --quick # Quick setup with defaults
- opal setup --profile # Use specific profile
- opal setup --show-systemd # Show systemd template
- opal setup --show-env # Show environment file template
Implementation:
- internal/wizard/prompts.go: Reusable prompt utilities
- internal/wizard/profiles.go: Profile definitions and templates
- cmd/setup.go: Main setup command implementation
- cmd/root.go: First-run detection and welcome message
- internal/engine/config.go: ConfigExists() and IsFirstRun() helpers
User experience:
- On first run, shows welcome message suggesting 'opal setup'
- Non-intrusive - creates defaults automatically if skipped
- Wizard guides through all configuration options
- Server setup includes OAuth/JWT configuration
- Environment file created with proper permissions (0600)
- Clear next steps displayed after completion
Separate configuration from data storage and make paths configurable
via environment variables and command-line flags. This improves
Unix/Linux compliance and supports both development and production
deployments.
Key changes:
- Separate config dir (opal.yml) from data dir (database, logs)
- Support XDG Base Directory specification
- Add --config-dir and --data-dir flags
- Environment variables: OPAL_CONFIG_DIR, OPAL_DATA_DIR, OPAL_DB_PATH
- Smart fallback: /etc/opal, /var/lib/opal -> ~/.config/opal, ~/.local/share/opal
- Server mode validates required OAuth/JWT environment variables
- Update naming from 'jade' to 'opal' throughout
- Update systemd service name to 'opal.service'
- Add migration guide in README
Default paths:
- Config: /etc/opal (fallback: ~/.config/opal)
- Data: /var/lib/opal (fallback: ~/.local/share/opal)
Files modified:
- internal/engine/config.go: New directory resolution logic
- internal/engine/database.go: Auto-create data directory
- cmd/root.go: Add global flags for directory overrides
- cmd/server.go: Add configuration validation
- cmd/sync.go, internal/sync/*: Use new path helper functions
- tests: Update to use directory overrides
- docs: Update deployment guide and README
- Create detailed deployment documentation
- Add Caddyfile.example with security headers and API proxy
- Document SystemD service setup for Go API
- Include database backup strategy
- Add troubleshooting guide
- Document OAuth configuration steps
- Add build and deployment commands
- Test production build successfully (340KB static site)
- Create TaskItem component with checkbox, meta info, tags
- Create TaskList component with loading/empty states
- Update home page with task list and filter toggle (pending/completed)
- Add new task form with description, project, priority, due date, tags
- Add task detail placeholder page
- Implement task toggle (complete/uncomplete)
- Add filter bar to switch between pending and completed tasks
- Support for optimistic UI updates and offline queueing
- Visual indicators for priority, due dates, overdue tasks
- Mobile-optimized list items with proper touch targets
- Add global CSS with mobile-first styling and CSS custom properties
- Create base UI components: Button, Input, Checkbox, Select
- Add BottomNav component with icons for Tasks, Projects, Tags, Settings
- Update app layout to include BottomNav and auth handling
- Create Settings page with API key input and sync controls
- Create auth pages: /auth/login (OAuth) and /auth/callback
- Add placeholder pages for Projects and Tags
- Implement manual API key authentication for MVP testing
- Add logout functionality and user info display
- Support for safe-area-inset (mobile notches)
- Create API client with auto-retry and token refresh support
- Add comprehensive API endpoints for tasks, tags, projects, sync, and auth
- Implement authStore for authentication state management
- Implement tasksStore with optimistic updates and offline queue
- Add derived stores for filtered task views (pending, completed, by project)
- Implement syncStore for managing sync state and queue
- Add client ID generation and persistence for sync tracking
- Add OAuth2 client for Authentik integration
- Implement JWT token generation and validation
- Add refresh token support with database storage
- Update database schema with oauth_subject, oauth_provider, and refresh_tokens table
- Create auth package with config, jwt, oauth, and token management
- Add OAuth endpoints: /auth/login, /auth/callback, /auth/refresh, /auth/logout
- Update AuthMiddleware to support both JWT and API key authentication
- Add user helper functions for OAuth user creation and retrieval
- Add .env.example with OAuth configuration template
API keys still work for CLI compatibility while JWT tokens support web/mobile clients.
BREAKING CHANGE: BuildWorkingSet() now accepts []*Task instead of *Filter
Problem:
- Working set IDs were assigned based on database query order (unsorted)
- Reports displayed tasks in sorted order (by urgency)
- Result: IDs didn't match displayed task positions (ID 1 wasn't first task)
Solution:
- Changed BuildWorkingSet() to accept pre-sorted task slice
- Reports now pass sorted tasks to BuildWorkingSet()
- IDs are assigned sequentially to match display order (1, 2, 3...)
Behavior:
- Reports rebuild working set on every execution with fresh IDs
- Task operations (done, modify, info) use saved working set IDs
- After completing a task, re-running report renumbers remaining tasks
Example:
Before: opal list shows ID 1 = low urgency task (wrong)
After: opal list shows ID 1 = highest urgency task (correct)
Tested scenarios:
✓ List report: IDs 1-N match urgency order
✓ Next report: IDs 1-5 match top urgent tasks
✓ Task completion: IDs renumber correctly after removal
✓ Multiple operations: Use saved working set (correct behavior)
✓ Different reports: Each builds own sequential IDs
- Add urgency calculation based on multiple factors:
* Due date (linear scale: overdue=12.0, today=10.0, week=6.0, 2weeks=2.0)
* Priority (H=6.0, M=3.9, D=1.8, L=0.0)
* Age (0-2.0 over 365 days)
* Active status (+4.0 boost)
* Waiting status (-3.0 penalty)
* Tags (+1.0 with count modifier)
* Project assignment (+1.0)
* Configurable urgent tag (default 'next', +15.0)
- Replace priority column with urgency in all reports
* Display as decimal with 1 decimal place
* 4-tier color coding: ≥10 (bright red), ≥5 (red), ≥2 (yellow), <2 (cyan)
* Minimal format color-coded by urgency
- Add default urgency sorting to all reports
* list, minimal, active, ready, overdue reports sort by urgency
* newest/oldest keep date-based sorting
- Implement 'next' report
* Shows most urgent ready tasks
* Configurable limit (default 5)
* Only includes tasks ready to work on (no future wait/scheduled)
- Add urgency display to info command
* Shows urgency score alongside priority
- All urgency coefficients configurable via config
* Adjusted defaults for Opal's simpler model (no blocking/annotations)
* Configurable urgent tag name (not hardcoded to 'next')
Priority order maintained: High > Medium > Default > Low
- Implement ProgressReporter interface with InteractiveProgress and NoOpProgress
- Add real-time progress bars using go-pretty/progress library
- Track 6 sync phases: connection test, queue push, pull, parse, apply, and server push
- Add --quiet flag to suppress progress output
- Auto-detect TTY to disable progress when piped/redirected
- Show task-level progress during apply phase with descriptions
- Display percentage complete and elapsed time for each phase
- Fix template task filtering bug: templates now hidden from all reports
except 'template' and 'all' reports, even when using custom filters
- Add support for status:template filter to explicitly show templates
- Implement comprehensive report system with 12 predefined reports:
* active - Started tasks
* all - All tasks including templates
* completed - Completed tasks
* list - Pending tasks (default)
* minimal - Pending tasks in minimal format
* newest - Most recent pending tasks
* oldest - Oldest pending tasks
* overdue - Overdue tasks
* ready - Tasks ready to work on
* recurring - Pending recurring instances
* template - Recurring template tasks
* waiting - Hidden/waiting tasks
- Replace list command with report-based architecture
- Add configurable default_report option (defaults to 'list')
- Add minimal display format (ID + description only)
- Support flexible syntax: 'opal <report> [filters]' or 'opal [filters] <report>'
- Add 'opal reports' command to list all available reports
Implement complete key:value format parsing for change log entries and fix
critical tag synchronization issue from server to client.
Key Changes:
1. Shared Key:Value Parser (NEW: internal/engine/parser.go)
- Created ParseKeyValueFormat() for both edit and sync operations
- Supports flexible whitespace: 'key:value' and 'key: value'
- Handles comment skipping for edit files
- Consolidates parsing logic (DRY principle)
2. Database Triggers - Tags Support (internal/engine/database.go)
- Added tags to track_task_create trigger
- Added tags to track_task_update trigger
- Tags sorted alphabetically via SQL ORDER BY
- Format: 'tags: alpha,bravo,charlie'
3. Task Creation - Tag Update Fix (internal/engine/task.go)
- CreateTaskWithModifier() now triggers update after adding tags
- Ensures tags appear in change log (UPDATE entry)
- Fixes missing tags in initial CREATE entries
4. Edit Command - Use Shared Parser (cmd/edit.go)
- Replaced custom parseEditedFile() with shared ParseKeyValueFormat()
- Added tag sorting in parseTags()
- Removed ~30 lines, improved maintainability
5. Sync Client - Complete Implementation (internal/sync/client.go)
- NEW: applyChangeDataToTask() - parses all fields from change log
- NEW: Helper functions for status, priority, tag parsing
- FIXED: parseChanges() - sort by timestamp+ID before grouping
- Added parent/child task ordering (prevents FK violations)
- Enhanced tag sync in merge loop with task reload
- Specific validation error messages per field
Critical Bug Fix:
- When CREATE and UPDATE have same timestamp, old code kept CREATE (no tags)
- New code sorts by ID as tiebreaker, ensuring UPDATE (with tags) is used
- Verified: Server->client tag sync now works correctly
Validation:
- Description must not be empty (both edit and sync)
- Recurrence validated (not negative, max 100 years)
- All timestamps parsed correctly (Unix epoch)
- Tags sorted alphabetically in all contexts
Testing:
- Fresh pull from server: ✅ All tags present
- API-created tasks: ✅ Tags sync correctly
- Local->server->client round-trip: ✅ No data loss
- Same-second CREATE+UPDATE: ✅ Correct entry processed
- Parent/child tasks: ✅ Correct ordering
Files Changed:
- NEW: internal/engine/parser.go (+44 lines)
- Modified: internal/engine/database.go (+10 lines)
- Modified: internal/engine/task.go (+8 lines)
- Modified: cmd/edit.go (-25 lines net)
- Modified: internal/sync/client.go (+280 lines)
- Modified: srv/README.md (+1 line)
Total: +318 lines added, -25 removed, net +293 lines
This completes Phase 6: Full bidirectional sync with complete tag support.
- Added 'opal sync merge' command for initial database merge
- Support for merge strategies: prefer-local, prefer-server, smart (default)
- Offline queue already implemented in Phase 2
- Conflict warning display already implemented in Phase 2-3
- Full offline mode support with automatic queueing when server unreachable
- Created comprehensive sync command suite:
- 'opal sync init' - Configure sync with server (URL, API key)
- 'opal sync status' - Show sync configuration and queue status
- 'opal sync now' - Bidirectional sync with conflict resolution
- 'opal sync up' - Push local changes to server
- 'opal sync down' - Pull server changes to local
- 'opal sync log' - View conflict resolution log
- Added interactive prompts for init (URL and API key)
- Automatic client ID generation (UUID)
- Display user-friendly sync results with emojis
- Support for viewing queued offline changes
- Integration with config system for persistent sync settings
- Created sync client for communicating with API server
- Implemented conflict resolution strategies (last-write-wins, server-wins, client-wins)
- Added offline change queue for queuing changes when server is unreachable
- Implemented merge logic for local and remote task lists
- Added conflict logging to sync_conflicts.log
- Created bidirectional sync with pull/push operations
- Extended Config struct with sync settings (URL, API key, client ID, strategy, offline queue)
- Added SyncResult display with user-friendly output
- Sync handlers already implemented in Phase 1 (GetChanges, PushChanges)
Replace manual string-based table formatting with jedib0t/go-pretty/v6/table
library to fix alignment issues with Norwegian characters (æøå) and ANSI
color codes.
Changes:
- Migrate FormatTaskList() to use go-pretty table
- Migrate FormatTaskDetail() to use clean table format
- Migrate FormatProjects() and FormatTagCounts() for consistency
- Remove truncate() function (no longer needed)
- Configure StyleLight with Unicode box-drawing characters
- Set proper column widths and alignment
Fixes:
- Priority column now center-aligned under 'Pri'
- Norwegian characters (æøå) display and align correctly
- Tags column properly aligned
- Description field truncates at 40 chars with proper UTF-8 handling
- All ANSI color codes handled automatically
- Consistent formatting across all table views
All existing color functions work unchanged. UTF-8 and ANSI codes are
handled automatically by go-pretty's width calculation.
When creating recurring tasks, modifiers (due, wait, priority, etc) were
not being applied because AttributeOrder was not copied to the temporary
modifier. This caused all date attributes to be ignored.
Refactored addRecurringTask to:
- Create task structs directly instead of using CreateTask (avoiding
premature saves)
- Use ApplyToNew() instead of Apply() for modifiers before first save
- Properly copy AttributeOrder when building the temporary modifier
- Save template and instance once with all fields correctly set
This ensures recurring tasks now properly have due dates, wait dates,
and other modifiers applied when created via 'opal add' or batch import.
joakim