_build_session_detail was omitting pending_task_lane, is_branching, and
active_branch_id from the GET /ai-sessions/{id} response. The fields
existed on the schema and model but were never passed in the manual
constructor, so task lane state could never be restored on navigation.
Also adds console logging to AssistantChatPage selectChat flow to
diagnose message restoration, and fixes ScriptTemplateEditor stepper
dismiss firing during programmatic script_body updates.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add PUT /ai-sessions/{id}/task-lane endpoint that saves the full task
lane state (AI questions/actions + user's in-progress responses) to
the pending_task_lane JSONB column. TaskLane debounce-saves to the
backend every 2s after changes. On session load, user responses are
restored from the backend into sessionStorage so TaskLane picks them
up on mount. Users can now close the browser, come back later, and
find their task lane exactly where they left it.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
PDF uploads were stored in S3 and had text extracted during upload, but
fetch_upload_images() filtered exclusively for image MIME types, so
document content never reached the AI.
- Add fetch_upload_documents() in storage_service.py to retrieve
extracted_content for PDFs and text files
- Update ai_sessions.py chat endpoint to call both fetch_upload_images
and fetch_upload_documents, injecting document text as context
- Add PDF text extraction in _generate_ai_description (pypdf)
- Add pypdf>=4.0.0 to requirements.txt
- Fix test_db teardown to avoid connection pool issues
- Add 5 tests for fetch_upload_documents
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix system prompt to ensure [QUESTIONS]/[ACTIONS] markers in AI responses
- Add format reminder injection to user messages for marker compliance
- Wire TaskLane activation in prefill and resume paths
- Add ActionCardGroup component for structured question/action rendering
- Update FlowPilot session and step card components
- Update ai-session schemas and types for marker data
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Integrates ResolutionOutputGenerator into the resolve endpoint so that
structured outputs (ticket note, KB article, etc.) are auto-generated
after every successful session resolution. Non-blocking — resolve still
succeeds if output generation fails.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The old /assistant/chats/* CRUD endpoints and assistant_chat_service
chat functions were unused — the frontend exclusively uses
/ai-sessions/{id}/chat (unified_chat_service) for all chat operations.
Removed:
- Chat CRUD endpoints (create, list, get, send, delete, conclude)
- assistant_chat_service: create_chat, send_message,
generate_conclusion_summary, CONCLUSION_SYSTEM_PROMPT
- Frontend: assistantChatApi chat methods, dead types
(AssistantChat, AssistantChatMessage, ConcludeChatRequest, etc.)
Kept:
- /assistant/retention endpoints (used by ChatRetentionSettingsPage)
- Shared AI infrastructure (_call_ai, _call_anthropic_cached,
ASSISTANT_SYSTEM_PROMPT, _auto_title) — imported by unified_chat_service
Moved:
- fetch_upload_images + resize_image_for_vision → storage_service.py
(shared location, not tied to dead endpoint)
Also added "Image Analysis" section to system prompt so Claude knows
to describe attached screenshots.
-650 lines of dead code removed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The frontend calls /ai-sessions/{id}/chat (unified_chat_service), not
/assistant/chats/{id}/messages (assistant_chat_service). The previous
commit wired images into the wrong backend. This fixes it:
- ai_session.py schema: add upload_ids to ChatMessageRequest
- ai_sessions.py endpoint: fetch images via _fetch_upload_images
- unified_chat_service: accept and forward images to _call_ai
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add session_type ('guided'|'chat') and title columns to ai_sessions,
enabling both FlowPilot guided sessions and assistant chat sessions to
live in a single table. This is the foundation for a unified session
history and consistent UX across both interaction modes.
Backend:
- Migration 066: session_type + title columns
- unified_chat_service: chat sessions on ai_sessions with same AI/RAG
- POST /ai-sessions supports session_type='chat' creation
- POST /ai-sessions/{id}/chat for chat messages
- DELETE /ai-sessions/{id} for session deletion
- session_type filter on GET /ai-sessions
Frontend:
- AssistantChatPage rewired to aiSessionsApi (no more assistantChatApi)
- /assistant/:sessionId route for deep-linking
- Session history: type filter pills (All/Guided/Chat), type icons
- Dashboard: both types shown with correct routing and icons
- Fixed glass-border → border-default in dashboard components
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Engineers can now generate AI-powered status updates during active FlowPilot
sessions and after resolve/escalate. Three audiences (Ticket Notes, Client
Update, Email Draft) with Quick/Detailed length options. Copy to clipboard
with one click. Client names auto-inserted from intake/PSA context.
Backend: new endpoint POST /ai-sessions/{id}/status-update with audience-aware
system prompts. Frontend: StatusUpdateModal with 2-step selection flow,
Share Update button in action bar, Share Resolution/Escalation on completed
sessions. Also updates Solutions Library spec with Community tier design.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Message bar now fixed-positioned above action bar with full-width
layout (respects both app sidebar and session sidebar)
- Added abandon_session endpoint (POST /ai-sessions/{id}/abandon)
- Added "Close" button to FlowPilot action bar with confirmation dialog
- Session can now be closed without resolving or escalating
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Users without team_id (solo accounts, pro plans) couldn't see
escalations because the query filtered by team_id which was NULL.
Now falls back to account_id scoping for both the escalation queue
endpoint and the sidebar badge count.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add escalation_count to sidebar stats (team-wide requesting_escalation)
- Show badge on Escalations nav item in sidebar
- Remove user_id filter from escalation queue — show all team escalations
including your own (needed for solo users and visibility)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The same model_validate bug existed in link_ticket endpoint (line 467).
Extracted the manual AISessionDetail construction into a shared helper
_build_session_detail() used by both get_session and link_ticket.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: AISessionDetail.model_validate(session) tried to validate the
ORM relationship 'steps' which has field 'id', but AISessionStepResponse
expects 'step_id'. This caused a Pydantic ValidationError on every session
detail load.
Fix: Construct AISessionDetail manually from ORM fields, passing the
already-built step_responses list directly instead of relying on
model_validate to serialize the ORM steps relationship.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Temporary debug commit — surfaces the actual exception when GET /ai-sessions/{id}
fails with 500, instead of generic "Internal server error" from middleware.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: embedding generation could break the DB transaction via a failed
SQL statement. The except block caught the Python error but left the transaction
in a failed state. Subsequent queries (_record_usage → subscription lookup)
then failed with InFailedSQLTransactionError.
Fixes:
- session_embedding_service: use begin_nested() savepoint so failures don't
poison the parent transaction
- ai_sessions.py: add db.rollback() before _record_usage in all 3 error
handlers (create, respond, pickup) to recover from broken transactions
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix AISession.ticket_id → psa_ticket_id in list_sessions filter query
- Add Gallery nav item (LayoutGrid icon) to AdminSidebar navItems array
- Remove ForeignKey from FileUpload.session_id (Python model) + migration b8d2f4a6c091 to drop DB constraint, allowing column to reference either session type
- Add 400ms debounce on AI session search input in SessionHistoryPage (aiSearchInput state + useRef timeout pattern)
- Show friendly 503 error message in RichTextInput upload error handler (both initial upload and retry paths)
- Add overflow-x-auto to FlowPilotAnalyticsPage tab bar container
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds vector-based similar session discovery using the existing Voyage AI
embedding infrastructure and pgvector cosine similarity search.
- New AISessionEmbedding model with vector(1024) column
- session_embedding_service: generate + upsert embeddings, find similar sessions
- Embeddings generated on session create (from problem_summary/domain) and
updated on resolve (adds resolution_summary)
- GET /ai-sessions/{id}/similar endpoint returns top-N similar sessions
- Migration a7c9e3b1f402 creates ai_session_embeddings table
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Fix escalation status mismatch: hook set 'escalated' but backend returns
'requesting_escalation'
- Fix list_sessions to include sessions picked up by Engineer B via
escalation_package->>'picked_up_by' JSONB query
- Fix sidebar escalation icon color: was Tailwind class 'text-amber-400'
passed to style={{color}}, now hex '#fbbf24'
- Replace window.location.reload() after ticket linking with
onReloadSession callback to preserve session state
- Update CLAUDE.md: Tailwind CSS v3 → v4 (@tailwindcss/vite, CSS-only config)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Implements Phase 1 of the FlowPilot-First pivot — the core AI session
experience where engineers describe a problem and FlowPilot guides them
through structured diagnosis with selectable options, free-text escape
hatches, and auto-generated documentation on resolution.
Backend: AISession + AISessionStep models, FlowPilot Engine (LLM
orchestration with structured JSON output), Flow Matching Engine v1
(semantic + keyword + recency scoring), 8 API endpoints with auth,
rate limiting, and AI quota enforcement.
Frontend: Intake screen, conversational session view with sidebar,
step cards with options/actions/resolution suggestions, resolve/escalate
modals, documentation view with rating, session history integration,
and /pilot route with sidebar navigation.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
