Implements three-phase AI assistant feature:
- Phase 0: RAG infrastructure with pgvector embeddings, Voyage AI integration,
tree chunking service, and semantic search over team's flow library
- Phase 1: In-session copilot panel during flow navigation with contextual
AI help, current step awareness, and suggested related flows
- Phase 2: Standalone AI chat page with persistent conversation history,
pin/delete, and configurable retention policies (account-level)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Apply glassmorphism styling, atmosphere orbs, branded wordmark, and
consistent design tokens to match the updated app shell aesthetic.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Updated Last Updated date, branding section (fonts, colors, glass
utilities, atmosphere orbs), component styling rules, and Design System
section to reflect the new ice-cyan glassmorphism theme.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Sweep of hardcoded purple hex values (#818cf8, #6366f1) replaced with
new cyan accent (#06b6d4) in QuickActions, RecentActivity, QuickLaunch,
and SVG brand assets.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New layout: greeting → calendar+actions → sessions+stats → activity
Replaces old QuickStats and SessionsPanel with new dashboard components
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Replace all color variables with Slate & Ice palette
- Add glass system vars (--glass-bg, --glass-blur, --shadow-float)
- Replace legacy glass-card with new variable-driven glass classes
- Add breatheGlow, bellWobble, slideDown, fadeInRight keyframes
- Update font references to IBM Plex Sans and Bricolage Grotesque
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Added Dockerfile.dev for backend and frontend, docker-compose.dev.yml
for local development. Added *.sql to .gitignore to prevent accidental
commit of database dumps.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Migrations must use sequential numbered prefixes (e.g., 040_desc.py) with
--rev-id flag. Never use auto-generated revision IDs. Updated both the
lessons learned section and the development commands.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Added lessons learned: AI tests need ai_enabled mock in CI (#28),
ESLint no-unused-vars lacks argsIgnorePattern (#29). Added Linux venv
activation path and gh CLI commands for CI debugging.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Frontend: suppress unused-vars ESLint errors for callback params in
MaintenanceFlowDetailPage and StepLibraryPage.
Backend: add autouse fixture to mock settings.ai_enabled=True in
test_ai_chat.py so tests pass in CI where no ANTHROPIC_API_KEY is set.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Updated from planning-phase placeholder to accurate representation of
ResolutionFlow as a production SaaS product. Reflects all shipped features
(AI Flow Assist, procedural/maintenance flows, cross-references, session
sharing, export improvements), current tech stack, project structure, and
development setup instructions.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Rewrote 03-DEVELOPMENT-ROADMAP.md — replaced outdated week-by-week format with
phase-based structure reflecting actual project state. Integrated all 13 open
GitHub issues into appropriate phases. Added "After Completing Work" checklist
to CLAUDE.md coding standards and roadmap/issues to quick reference table.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- AI conversational flow builder with interview phases (scoping, discovery, enrichment)
- Flow Assist sidebar nav, toolbar, chat panel, tree preview
- Cross-reference / loop-back support: nodes can reference any node in the tree
- Dashed purple arrows on canvas for cross-references
- Node picker dropdowns for action and decision forms
- Backend validator relaxed for cross-references
- AI system prompt updated for loop-back patterns
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Design for allowing tree nodes to reference any node in the tree
(not just direct children), enabling loop-back patterns. Ghost
reference approach — no schema changes, SVG overlay arrows on
canvas, node picker + drag-to-link in editor.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- After generation, toolbar shows "Save to Flow Library" button
(replaces "Import to Editor")
- Button shows "Saving..." spinner state during API call
- Generate button shows animated spinner during generation
- Backend /import endpoint always creates a new Tree record
(removed generated_tree_id idempotency check)
- Navigates to tree editor after successful save
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Rename all "Build with AI" labels to "Flow Assist"
- Update CreateFlowDropdown description to "AI-powered flow builder"
- Add Flow Assist nav item with Sparkles icon to sidebar (both
expanded and collapsed states)
- Update AIFlowBuilderModal titles for consistency
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Strip unclosed [TREE_UPDATE] and [METADATA] blocks from display when
response is truncated by max_tokens
- Increase send_message max_tokens from 2000 to 8000 to prevent
truncation of large tree JSON
- Use lightweight validation for progressive tree updates (valid root
node only) instead of strict 5-node minimum — strict validation
still applies at final /generate step
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- C1: Fix race condition in handleReset — await abandonSession before
starting new session to prevent store state corruption
- I1: Extract error messages from Axios response.data.detail instead of
generic error.message — users now see helpful backend messages (quota
limits, message caps, etc.)
- I2: Add isGenerating guard in generateTree store action to prevent
concurrent generation requests on double-click
- I3: Add isResponding guard in sendMessage to prevent concurrent sends
- M5: Remove redundant type casts on flowType
- M6: Add rate limiter to DELETE /sessions/{id} for consistency
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- TypeScript types for chat session, messages, and responses
- API client module with all 6 endpoints
- Zustand store with session management, message sending, tree generation, import, resume
- 7 chat components: ChatMessage, ChatInput, ChatPanel, PhaseIndicator, ChatToolbar, EmptyPreview, StaticTreePreview
- AIChatBuilderPage with split-panel layout (60% chat / 40% preview)
- Route at /ai/chat with lazy loading
- "Build with AI" button on TreeLibraryPage
- Session resume via URL search params
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tests cover session create, send message with tree update, get session,
abandon, 404 on missing session, and 503 when AI disabled.
Fixed: ai_usage.conversation_id has FK to ai_conversations, not
ai_chat_sessions. Chat builder now passes conversation_id=None and
tracks session reference in extra_data.chat_session_id.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
6 endpoints: create session, send message, get session, generate tree,
import to editor, abandon. Quota service daily limit updated to include
chat builder generation types.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The AI Chat Builder needs conversational text responses, not JSON-only.
Gemini's generate_json forces response_mime_type='application/json'
which is incompatible. The new generate_text method omits this constraint.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
14 tasks across 4 phases: backend foundation (model, schemas, service,
endpoints, tests), frontend chat UI (types, API client, store,
components, page), tree preview integration, and polish (session
resume, responsive layout).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Design for conversational AI flow builder that interviews users as a
senior MSP engineer to collaboratively build troubleshooting/procedural
flows. Coexists with existing wizard-based builder as a separate entry
point.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The "Unterminated string" JSON parse error is likely caused by Gemini
output truncation at 1024 tokens. Increases scaffold max_tokens to 2048
and adds logging for: raw response text, finish_reason (truncation
detection), and JSON parse failures.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The google-genai SDK uses `client.aio.models.generate_content()` for
async calls, not `client.models.generate_content_async()` which doesn't
exist. Also removes the temporary /ai/provider-debug endpoint.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Shows which provider is selected and whether keys are loaded.
Remove after debugging the 502 on PR env.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
