Opslaan faalde altijd door label/code-mismatch met de database-constraints
(reviewbevinding). Tabs komen terug bij de rebuild op het nieuwe datamodel.
- tab-mappen anamnese/, examination/, rom/ verwijderd
- tabs uit intake-tabs.tsx, server actions uit actions.ts
- intake-status API telt de secties niet meer mee (voortgangsring klopt)
- Cortex-navigatiedoelen (regex, entity-mapping, prompt) bijgewerkt zodat
'ga naar anamnese' geen 404 meer geeft
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Cortex handelt een no-show af vanuit één chatcommando: afspraak
annuleren (declarabiliteits-nudge), en de openstaande concept-
huisartsbrief wordt via LLM herschreven en ter review aangeboden
in een document artifact (human-in-the-loop, PATCH dispatch zet
status op verzendklaar).
- API-routes: context, rescript, cancel, dispatch
- NoShowDocumentBlock: review/edit UI met origineel-vergelijk
- Mock-data voor concept huisartsbrief
- PRD, FO, bouwplan en epics in docs/intent/noshow-case/
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Fixes 4 bugs preventing intake blocks from working:
1. Chat API validation: Add 'nudge' to message type enum, allow empty
content for streaming messages
2. AI chat recognition: Add P3 intake intents (intake_status,
risico_query, diagnose_query, intake_navigeer) to system prompt
with triggers, entities, and JSON examples
3. Artifact rendering: Add intake block imports and switch cases
to artifact-container.tsx
4. API 400 error: Convert null to undefined for optional Zod params
(searchParams.get returns null, Zod .optional() expects undefined)
Also adds:
- Testplan for E2E testing (60+ test cases)
- Session log with lessons learned and new intent checklist
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Adds 4 new intents for intake workflow:
- intake_status: shows completion percentage and section checklist
- intake_navigeer: navigation to specific intake tabs
- risico_query: displays risk assessments with severity indicators
- diagnose_query: shows primary/secondary diagnoses
New components:
- IntakeStatusBlock, RisicoBlock, DiagnoseBlock
- Shared block components (BlockLoading, BlockError, BlockEmpty, BlockSection, BlockItem, BlockFooter)
- useBlockData and useIntakeContext hooks
New API routes:
- GET /api/cortex/intake/status
- GET /api/cortex/intake/risico
- GET /api/cortex/intake/diagnose
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
System prompt aangepast voor nuchtere, understated tone of voice:
- Kernkwaliteit "Nuchter" toegevoegd
- Tone of voice: droog en behulpzaam ipv vriendelijk
- Stijlvoorbeelden sectie met concrete frases
- 6 voorbeeldresponses aangepast naar kortere, drogere stijl
- Error messages menselijker ("Hmm, snap niet helemaal...")
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Voornaam zoeken werkte niet omdat name_given array case-sensitive is.
Capitalize nu de zoekterm (eerste letter hoofdletter) voordat deze
wordt gebruikt in de array search, zodat 'jan' matcht met 'Jan'.
Aangepast in beide search endpoints:
- /api/patients/search
- /api/cortex/patients/search
- Adjusted date handling in the agenda API to ensure optional parameters for start, end, and label are correctly processed.
- Enhanced chat input and action parser to support optional date labels for relative dates.
- Updated agenda components to handle date ranges and loading states more effectively.
- Improved error handling and loading indicators in the agenda block for better user experience.
This commit ensures consistency in date handling across the application, aligning with the new requirements for relative date inputs.
SOC: Parser en entity-extractor accepteren nu optionele referenceDate
parameter. Caller bepaalt wat "nu" is, niet de functies intern.
Wijzigingen:
- parseRelativeDate(input, referenceDate?) - referenceDate parameter
- extractEntities(input, intent, referenceDate?) - doorgeeft aan parser
- API routes gebruiken new Date() als referenceDate
Verwijderd (YAGNI):
- lib/utils/server-date.ts - wrapper om new Date() was overbodig
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implements AI-powered multi-intent classification for Cortex V2.
When Reflex Arc (Layer 1) detects complexity, it escalates to
the Orchestrator for intelligent classification using Claude 3.5 Haiku.
E2.S1 - System prompt for Orchestrator
- ORCHESTRATOR_SYSTEM_PROMPT constant with Dutch instructions
- Multi-intent detection (en, daarna, ook, eerst, vervolgens)
- Pronoun resolution rules (hij/zij → active patient)
- JSON-only output format specification
E2.S2 - Context formatting for AI
- formatContextForPrompt() with emoji structure
- Dutch labels for shifts and views
- Graceful null handling, token limiting (max 5 agenda, 3 recent)
E2.S3 - AI classification endpoint
- classifyWithOrchestrator() calling Claude 3.5 Haiku
- Builds IntentChain from AI response
- Tracks tokens used and processing time
E2.S4 - IntentChain parsing with Zod
- parseAIResponse() with Zod schema validation
- Strips markdown code blocks
- Partial extraction for malformed responses
- Graceful fallback to intent: 'unknown'
E2.S5 - POST /api/cortex/classify hybrid endpoint
- Reflex first, escalate to Orchestrator if needed
- Response includes handledBy: 'reflex' | 'orchestrator'
- Auth check, feature flag check, request validation
E2.S6 - Graceful fallback
- buildFallbackChain() with confidence cap (0.6)
- buildChainFromReflex() for simple chains
- classifyWithTimeout() wrapper (5s default)
New files:
- lib/cortex/orchestrator.ts (~780 lines)
- app/api/cortex/classify/route.ts (~180 lines)
Key deliverable: "Zeg Jan af en maak notitie" correctly
parsed into 2 actions in an IntentChain.
Progress: 27 SP / 48 SP (56%) - E0, E1, E2 complete
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
