Replace all inline session management with the shared useAssistantSession
hook. Keep cockpit-specific state (triageMeta, workZonePct, steps, onboarding)
and handlers. Wire onSessionLoadedRef/onTriageUpdateRef callbacks. Add feature
flag redirect for flowpilot_cockpit. Update router and prefetch references.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Recreates the production AssistantChatPage as FlowPilotPage using the
shared useAssistantSession hook. Classic chat interface with ChatMessage
bubbles, TaskLane side panel, rich input with file uploads, and
conclude/status update modals. ViewToggle commented out pending Task 4.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
No external call sites existed. Feature gating now handled by
useFeatureFlag hook backed by the feature_flags system.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Selector-based hook reads from authStore.featureFlags.
Returns false for unknown keys (fail closed).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Loads resolved feature flags from /auth/me/feature-flags alongside
user, account, and subscription data.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Resolves feature flags for the current user using:
account override > plan default > false
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
handleSend, sendPrefill, and handleResumeNew all make async API calls
that can return after the user has switched to a different session. Without
a guard, the stale response overwrites the new session's questions/actions
state — causing the previous session's FlowPilot Asks to persist.
Fix: capture the session ID before each await and check currentChatRef
after — discarding the response if the user has since switched. This
matches the existing guard pattern in selectChat (lesson #106).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Stacked zone layout: incident header → work zone → drag handle →
conversation log → compose
- IncidentHeader wired with triageMeta state and field save handlers
- Work zone: StepsPanel (left) + FlowPilotAsks + WhatWeKnow (right)
- Drag-resizable split with localStorage persistence
- Compact conversation log with you:/fp: prefixes
- Triage state populated on session load/resume
- AI triage_update merged into header via mergeTriageUpdate()
- MSP-native language: FlowPilot, New Case, Close Case
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- IncidentHeader: labelled fields with per-field edit popovers
- StepsPanel: ordered step checklist (✓/→/○) with script CTA
- FlowPilotAsks: quick-reply buttons or free-text input
- WhatWeKnow: evidence list with status toggle and inline editing
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- TriageMeta, EvidenceItem, TriageUpdate interfaces
- QuestionItem.options field for quick-reply buttons
- triage_update on ChatMessageResponse
- Triage fields on AISessionDetail for session resume
- updateTriage() and getHandoffDraft() API methods
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add _parse_triage_update_marker() parser following existing marker pattern
- Add [TRIAGE_UPDATE] instructions to system prompt with grounding rules
- Add QuestionItem.options support in question parser
- Wire triage extraction into both main and branch-aware chat paths
- Auto-PATCH session: AI only fills null fields (manual edits win)
- Evidence items: AI appends only, never modifies existing
- Return triage_update in ChatMessageResponse for frontend header sync
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Addresses every Red and Yellow item from the codex review:
- Canonical handoff: ResolutionOutputGenerator is the source of truth
- AI vs manual authority: manual edits win, AI never overwrites
- evidence_items: full-list replacement, frontend is merge authority
- TaskLane persistence: lifted into hook, StepsPanel is presentation-only
- Quick replies: immediate-send, full-stack contract change
- issue_category + asset_name: free text in v1
- Adds 5 implementation guardrails and Phase 2 gate for triage extraction
- Execution order updated to 37 steps with persistence extraction step
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Merges MSP_Assistant_Harness_Implementation_Plan.docx with the
brainstorming design spec into a single executable plan. Resolves
all open questions from the original docx, expands scope to include
backend changes, and adds a 35-step phased execution order.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Design spec for evolving /assistant into a live triage cockpit.
Covers layout decisions (stacked zones, drag-resizable split),
incident header (labelled fields, AI-inferred + editable),
work zone (steps checklist + FlowPilot Asks + What We Know),
conclude modal redesign, and all required backend changes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
#60a5fa (electric blue) is a light colour — near-white and muted-foreground
text on it produced ~1.9–2.1:1 contrast, well below WCAG AA.
- TreeNavigationPage: option number badges + timer badge → text-[#0e1016]
- StepChecklist: active step was blue badge on blue row (invisible); now
inverts to navy circle with blue number when current, dark-on-blue otherwise
- StepDetail: step number circle → text-[#0e1016] (was near-white on light blue)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Added 'All Scripts' as the first/default tab (no mine/shared filter) so the
page opens with every script the user can access. My Scripts and Team Scripts
tabs remain for filtered views.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
setCategory and setSearch called loadTemplates() with no arguments, dropping
the mine/shared filter set by the active tab — causing the list to show all
templates instead of the current user's scripts after any filter interaction.
Store now persists tabFilters whenever loadTemplates is called with explicit
filters, and reuses them for subsequent no-arg calls from setCategory/setSearch.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- script_builder endpoint: pg_advisory_xact_lock on user_id before
session count check, preventing concurrent creates from both passing
the MAX_SESSIONS_PER_USER guard
- script_builder_service send_message: pg_advisory_xact_lock on session_id
before message count check, preventing concurrent sends from both
passing the MAX_MESSAGES_PER_SESSION guard
- script_builder_service save_to_library: replace check-then-insert slug
logic with IntegrityError retry loop (3 attempts with fresh UUID suffix);
add unique constraint on script_templates.slug (migration 070)
- ScriptBuilderPage: add creatingSessionRef to serialize concurrent
handleSend calls that would otherwise both call createSession() while
session is still null
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Race condition: on page remount, selectChat(oldId) loads session data async.
If the user clicks New Chat before the API returns, the old session's
pending_task_lane was being applied to the new session's state, showing
stale tasks and blocking new ones from appearing.
Fix: currentChatRef tracks the most recently requested chat ID synchronously.
All chat-creation paths (selectChat, handleNewChat, handleResumeNew) update it
immediately. After each await in selectChat, bail if the ref no longer matches.
Also documents the pattern as Lesson 106 in CLAUDE.md for future reference.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
VITE_SENTRY_DSN is already set in Railway as a build arg. The hardcoded
fallback was unnecessary and triggered GitHub secret scanning alerts.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- landing.css: hardcode --lp-btn to #60a5fa (lesson 104 — no var(--color-*) in landing.css)
- ScriptBuilderInput: suggestion chips now correctly disabled during generation
- ChatSidebar: wrapper onClick no longer fires onSelect while in confirming state
- SessionHistoryPage: fix loadMoreAiSessions race condition with generation counter;
flow session tab auto-activates when URL params target flow session filters
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
