## DevLoop - Closed-Loop Development System
- Orchestrator for plan → build → test → analyze cycle
- Gemini planning via OpenCode CLI
- Claude implementation via CLI bridge
- Playwright browser testing integration
- Test runner with API, filesystem, and browser tests
- Persistent state in .devloop/ directory
- CLI tool: tools/devloop_cli.py
Usage:
python tools/devloop_cli.py start 'Create new feature'
python tools/devloop_cli.py plan 'Fix bug in X'
python tools/devloop_cli.py test --study support_arm
python tools/devloop_cli.py browser --level full
## HTML Reports (optimization_engine/reporting/)
- Interactive Plotly-based reports
- Convergence plot, Pareto front, parallel coordinates
- Parameter importance analysis
- Self-contained HTML (offline-capable)
- Tailwind CSS styling
## Playwright E2E Tests
- Home page tests
- Test results in test-results/
## LAC Knowledge Base Updates
- Session insights (failures, workarounds, patterns)
- Optimization memory for arm support study
Phase 4 - Live Updates:
- Create useOptimizationStream hook for real-time trial updates
- Replace polling with WebSocket subscription in SpecRenderer
- Auto-report errors to ErrorPanel via panel store
- Add progress tracking (FEA count, NN count, best trial)
Phase 5 - Convergence Visualization:
- Add ConvergenceSparkline component for mini line charts
- Add ProgressRing component for circular progress indicator
- Update ObjectiveNode to show convergence trend sparkline
- Add history field to ObjectiveNodeData schema
- Add live progress indicator centered on canvas when running
Bug fixes:
- Fix TypeScript errors in FloatingIntrospectionPanel (type casts)
- Fix ValidationPanel using wrong store method (selectNode vs setSelectedNodeId)
- Fix NodeConfigPanelV2 unused state variable
- Fix specValidator source.extractor_id path
- Clean up unused imports across components
Phase 1 - Panel Management System:
- Create usePanelStore.ts for centralized panel state management
- Add PanelContainer.tsx for draggable floating panels
- Create FloatingIntrospectionPanel.tsx (persistent, doesn't disappear on node click)
- Create ResultsPanel.tsx for trial result details
- Refactor NodeConfigPanelV2 to use panel store for introspection
- Integrate PanelContainer into CanvasView
Phase 2 - Pre-run Validation:
- Create specValidator.ts with comprehensive validation rules
- Add ValidationPanel (enhanced version with error navigation)
- Add Validate button to SpecRenderer with status indicator
- Block run if validation fails
- Check for: design vars, objectives, extractors, bounds, connections
Phase 3 - Error Handling & Recovery:
- Create ErrorPanel.tsx for displaying optimization errors
- Add error classification (nx_crash, solver_fail, extractor_error, etc.)
- Add recovery suggestions based on error type
- Update status endpoint to return error info
- Add _get_study_error_info helper to check error_status.json and DB
- Integrate error detection into status polling
Documentation:
- Add CANVAS_ROBUSTNESS_PLAN.md with full implementation plan
Phase 2 - Execution Bridge:
- Update /start endpoint to fallback to generic runner when no study script exists
- Auto-detect model files (.prt, .sim) from 1_setup/model/ directory
- Pass atomizer_spec.json path to generic runner
Phase 3 - Live Monitoring & Results Overlay:
- Add ResultBadge component for displaying values on canvas nodes
- Extend schema with resultValue and isFeasible fields
- Update DesignVarNode, ObjectiveNode, ConstraintNode, ExtractorNode to show results
- Add Run/Stop buttons and Results toggle to SpecRenderer
- Poll /status endpoint every 3s and map best_trial values to nodes
- Show green/red badges for constraint feasibility
- Create introspect_sim.py NX journal to extract solutions, BCs from SIM files
- Update introspect_sim_file() to optionally call NX journal for full introspection
- Add FEM mesh section (nodes, elements, materials, properties) to IntrospectionPanel
- Add SIM solutions and boundary conditions sections to IntrospectionPanel
- Show introspection method and NX errors in panel
- Fetch available parts from /nx/parts on panel mount
- Dropdown to select which part to introspect (default = assembly)
- Hides idealized parts (*_i.prt) from dropdown
- Shows part size in dropdown (KB or MB)
- Header shows selected part name in primary color
- Refresh button respects current part selection
- Auto-introspects when part selection changes
Backend:
- GET /nx/parts - List all .prt files in model directory
- GET /nx/introspect/{part_name} - Introspect a specific part file
(e.g., M1_Blank.prt instead of just the assembly)
- Each part gets its own cache file (_introspection_{stem}.json)
Frontend IntrospectionPanel:
- Add 'FEA Results' section showing existing OP2/F06 sources
- Green checkmark when results exist, shows recommended source
- Expand file_deps and fea_results sections by default
- Add CheckCircle2 and Database icons
This allows introspecting component parts that contain the actual
design variable expressions (e.g., M1_Blank has 56 expressions
while the assembly ASSY_M1 only has 5).
- scan_existing_fea_results() scans study for existing OP2/F06 files
- Introspection now returns existing_fea_results with recommended source
- New POST /nx/extract endpoint runs extractors on existing OP2 files
- Supports: displacement, stress, frequency, mass_bdf, zernike
- No NX solve needed - uses PyNastran and Atomizer extractors directly
This allows users to test extractors and get physics data from existing
simulation results without re-running the FEA solver.
- Use same approach as run_optimization.py with use_iteration_folders=True
- NXSolver.create_iteration_folder() handles proper file copying
- Read NX settings from atomizer_spec.json or optimization_config.json
- Extract Nastran version from NX install path
- Creates iter0 folder for baseline (consistent with optimization numbering)
This fixes the issue where manually copying files didn't preserve
NX file dependency chain (.sim -> .afm -> .fem -> _i.prt -> .prt)
- Parse Nastran log file to detect memory allocation failures
- Extract requested vs available memory from log
- Provide actionable error message with specific values
- Include log files in result_files response
- Read nx_install_path from atomizer_spec.json if available
- Auto-detect from common Siemens installation paths
- Fixes issue where NX2512 wasn't found (actual path is DesigncenterNX2512)
Drag-drop fixes:
- Fix Objective default data: use nested 'source' object with extractor_id/output_name
- Fix Constraint default data: use 'type' field (not constraint_type), 'threshold' (not limit)
Undo/Redo fixes:
- Remove dependency on isDirty flag (which is always false due to auto-save)
- Record snapshots based on actual spec changes via deep comparison
Code generation improvements:
- Update system prompt to support multiple extractor types:
* OP2-based extractors for FEA results (stress, displacement, frequency)
* Expression-based extractors for NX model values (dimensions, volumes)
* Computed extractors for derived values (no FEA needed)
- Claude will now choose appropriate signature based on user's description
Backend:
- Add POST /api/optimization/studies/{study_id}/nx/run-baseline endpoint
- Creates trial_baseline folder in 2_iterations/
- Copies all model files and runs NXSolver
- Returns paths to result files (.op2, .f06, .bdf) for extractor testing
Frontend:
- Add 'Run Baseline Simulation' button to IntrospectionPanel
- Show progress spinner during simulation
- Display result files when complete (OP2, F06, BDF)
- Show error messages if simulation fails
This enables:
- Testing custom extractors against real FEA results
- Validating the simulation pipeline before optimization
- Inspecting boundary conditions and loads
CanvasView:
- Fix Save button visibility - now shows when spec is loaded (grayed if no changes)
- Separate logic for spec mode vs legacy mode save buttons
- Fix Reload button visibility
IntrospectionPanel:
- Add Mass Properties section (mass, volume, surface area, CoG, body count)
- Add Linked Parts section showing file dependencies
- Add Bodies section (solid/sheet body counts)
- Add Units section showing unit system
- Type-safe access to all nested properties
- Handle expressions as object with user/internal arrays (new format) or direct array (old)
- Add useMemo for expression filtering
- Make extractors_available, dependent_files, warnings optional with safe access
- Support both 'unit' and 'units' field names
- SpecRenderer: Add localNodes state with applyNodeChanges for smooth node dragging
- SpecRenderer: Fix getDefaultNodeData() - extractor uses 'custom_function' type with function definition
- SpecRenderer: Fix constraint default - use constraint_type instead of type
- CanvasView: Show config panel INSTEAD of chat when node selected (not blocked)
- NodeConfigPanelV2: Enable showHeader for code editor toolbar (Generate/Snippets/Validate/Test buttons)
- NodeConfigPanelV2: Pass studyId to IntrospectionPanel
- IntrospectionPanel: Accept studyId prop and use correct API endpoint
- optimization.py: Search multiple directories for model files including 1_setup/model/
Backend:
- Add POST /generate-extractor for AI code generation via Claude CLI
- Add POST /generate-extractor/stream for SSE streaming generation
- Add POST /validate-extractor with enhanced syntax checking
- Add POST /check-dependencies for import analysis
- Add POST /test-extractor for live OP2 file testing
- Add ClaudeCodeSession service for managing CLI sessions
Frontend:
- Add lib/api/claude.ts with typed API functions
- Enhance CodeEditorPanel with:
- Streaming generation with live preview
- Code snippets library (6 templates: displacement, stress, frequency, mass, energy, reaction)
- Test button for live OP2 validation
- Cancel button for stopping generation
- Dependency warnings display
- Integrate streaming and testing into NodeConfigPanelV2
Uses Claude CLI (--print mode) to leverage Pro/Max subscription without API costs.
Phase 1 - Foundation:
- Add NodeConfigPanelV2 using useSpecStore for AtomizerSpec v2.0 mode
- Deprecate AtomizerCanvas and useCanvasStore with migration docs
- Add VITE_USE_LEGACY_CANVAS env var for emergency fallback
- Enhance NodePalette with collapse support, filtering, exports
- Add drag-drop support to SpecRenderer with default node data
- Setup test infrastructure (Vitest + Playwright configs)
- Add useSpecStore unit tests (15 tests)
Phase 2 - File Structure & Model:
- Create FileStructurePanel with tree view of study files
- Add ModelNodeV2 with collapsible file dependencies
- Add tabbed left sidebar (Components/Files tabs)
- Add GET /api/files/structure/{study_id} backend endpoint
- Auto-expand 1_setup folders in file tree
- Show model file introspection with solver type and expressions
Technical:
- All TypeScript checks pass
- All 15 unit tests pass
- Production build successful
## Documentation Updates
- DASHBOARD.md: Updated to V3.0 with Canvas V3 features, file browser, introspection
- DASHBOARD_IMPLEMENTATION_STATUS.md: Marked Canvas V3 features as COMPLETE
- CANVAS.md: New comprehensive guide for Canvas Builder V3 with all features
- CLAUDE.md: Added dashboard quick reference and Canvas V3 features
## Canvas V3 Features Documented
- File Browser: Browse studies directory for model files
- Model Introspection: Auto-discover expressions, solver type, dependencies
- One-Click Add: Add expressions as design variables instantly
- Claude Bug Fixes: WebSocket reconnection, SQL errors resolved
- Health Check: /api/health endpoint for monitoring
## Backend Services
- NX introspection service with expression discovery
- File browser API with type filtering
- Claude session management improvements
- Context builder enhancements
## Frontend Components
- FileBrowser: Modal for file selection with search
- IntrospectionPanel: View discovered model information
- ExpressionSelector: Dropdown for design variable configuration
- Improved chat hooks with reconnection logic
## Plan Documents
- Added RALPH_LOOP_CANVAS_V2/V3 implementation records
- Added ATOMIZER_DASHBOARD_V2_MASTER_PLAN
- Added investigation and sync documentation
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 1-7 of Canvas V4 Ralph Loop implementation:
Backend:
- Add /api/files routes for browsing model files
- Add /api/nx routes for NX model introspection
- Add NXIntrospector service to discover expressions and extractors
- Add health check with database status
Frontend:
- Add FileBrowser component for selecting .sim/.prt/.fem files
- Add IntrospectionPanel to discover expressions and extractors
- Update NodeConfigPanel with browse and introspect buttons
- Update schema with NODE_HANDLES for proper flow direction
- Update validation for correct DesignVar -> Model -> Solver flow
- Update useCanvasStore.addNode() to accept custom data
Flow correction: Design Variables now connect TO Model (as source),
not FROM Model. This matches the actual data flow in optimization.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Bug Fixes:
- Fix Atomizer Assistant error with reconnect button and error state handling
- Enable connection/edge deletion with keyboard Delete/Backspace keys
- Fix drag & drop positioning using screenToFlowPosition correctly
- Fix loadFromConfig to create all node types and edges properly
UI/UX Improvements:
- Minimal responsive header with context breadcrumb
- Better contrast with white text on dark backgrounds
- Larger font sizes in NodePalette for readability
- Study-aware header showing selected study name
New Features:
- Enhanced ExecuteDialog with Create/Update mode toggle
- Select existing study to update or create new study
- Home page Canvas Builder button for quick access
- Home navigation button in CanvasView header
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Ensure MCP config file is created before CLI calls
- Pass --mcp-config flag to Claude CLI for atomizer tools access
- Add mcp__atomizer-tools__* to allowed tools in user mode
- Enables canvas validate/execute/interpret intent tools
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add textarea input with Enter-to-send functionality to ChatPanel
- Pass sendMessage and isConnected props from AtomizerCanvas
- Add disabled states during thinking/disconnected states
- Use Lucide Send icon for submit button
- Show connection status with Sparkles welcome icon
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 5 of Canvas Professional Upgrade:
- Create ExpressionSelector component with searchable dropdown
- Fetch expressions from NX introspection API
- Auto-populate label, units, and default min/max from expression value
- Add refresh button to reload expressions
- Integrate into NodeConfigPanel for DesignVar nodes
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 4 of Canvas Professional Upgrade:
- Add /studies/{id}/nx/introspect endpoint for full model introspection
- Add /studies/{id}/nx/expressions endpoint for expression list
- Add caching to avoid re-running NX journal on each request
- Add frontend API client methods: introspectNxModel, getNxExpressions
- Use existing introspect_part.py extractor
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 3 of Canvas Professional Upgrade:
- Fix ConfigImporter API endpoint URL (/api/optimization/studies/...)
- Add currentStudyId prop to auto-select study in importer
- Parse response correctly (data.config instead of raw response)
- Replace file upload emoji with FolderOpen Lucide icon
- Add "Load Study Config" button to Setup canvas tab header
- Show button only when canvas is empty
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 2 of Canvas Professional Upgrade:
- Update AtomizerCanvas with Lucide icons (MessageCircle, Plug, X)
- Update CanvasView page with dark theme styling
- Replace emoji buttons with Lucide icons (ClipboardList, Download, Trash2)
- Update Setup canvas tab for full-width responsive layout
- Use calc(100vh-6rem) for proper canvas height
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Phase 1 of Canvas Professional Upgrade:
- Replace all emoji icons in nodes with Lucide components
- Update BaseNode to use iconColor prop instead of color/colorBg
- Update NodePalette with matching Lucide icons
- Update templates.ts with Lucide icon names (Box, Scale, BarChart3, etc.)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Canvas Builder Visual Updates:
- Update all Canvas components to use Atomaster dark theme
- BaseNode: dark background (bg-dark-800), white text, primary selection glow
- NodePalette: dark sidebar with hover states
- NodeConfigPanel: dark inputs, labels, and panel background
- ValidationPanel: semi-transparent error/warning panels with backdrop blur
- ChatPanel: dark message area with themed welcome state
- ExecuteDialog: dark modal with primary button styling
- ConfigImporter: dark tabs, inputs, and file upload zone
- TemplateSelector: dark cards with category pills and hover effects
Setup Page Integration:
- Add Configuration/Canvas Builder tab switcher
- Canvas tab renders AtomizerCanvas full-height
- Tabs styled to match Atomaster theme
Build: Passes TypeScript and Vite build
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add canvas.ts MCP tool with validate_canvas_intent, execute_canvas_intent, interpret_canvas_intent
- Add useCanvasChat.ts bridge hook connecting canvas to chat system
- Update context_builder.py with canvas tool instructions
- Add ExecuteDialog for study name input
- Add ChatPanel for canvas-integrated Claude responses
- Connect AtomizerCanvas to Claude via useCanvasChat
Canvas workflow now:
1. Build graph visually
2. Click Validate/Analyze/Execute
3. Claude processes intent via MCP tools
4. Response shown in integrated chat panel
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Major changes:
- Dashboard: WebSocket-based chat with session management
- Dashboard: New chat components (ChatPane, ChatInput, ModeToggle)
- Dashboard: Enhanced UI with parallel coordinates chart
- MCP Server: New atomizer-tools server for Claude integration
- Extractors: Enhanced Zernike OPD extractor
- Reports: Improved report generator
New studies (configs and scripts only):
- M1 Mirror: Cost reduction campaign studies
- Simple Beam, Simple Bracket, UAV Arm studies
Note: Large iteration data (2_iterations/, best_design_archive/)
excluded via .gitignore - kept on local Gitea only.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
