feat(web): operations-first control plane UI revamp with live updates and node logs#330
Merged
santoshkumarradha merged 114 commits intomainfrom Apr 6, 2026
Merged
Conversation
Replace all @phosphor-icons/react imports with lucide-react equivalents. Rewrote icon-bridge.tsx to re-export Lucide icons under the same names used throughout the codebase, so no consumer files needed changing. Updated icon.tsx to use Lucide directly. Removed weight= props from badge.tsx, segmented-status-filter.tsx, and ReasonerCard.tsx since Lucide does not support the Phosphor weight API.
…tures being redesigned - Delete src/components/mcp/ (MCPServerList, MCPServerCard, MCPHealthIndicator, MCPServerControls, MCPToolExplorer, MCPToolTester) - Delete src/components/authorization/ (AccessRulesTab, AgentTagsTab, ApproveWithContextDialog, PolicyContextPanel, PolicyFormDialog, RevokeDialog) - Delete src/components/packages/ (AgentPackageCard, AgentPackageList) - Delete src/components/did/ (DIDIdentityCard, DIDDisplay, DIDStatusBadge, DIDInfoModal, DIDIndicator) - Delete src/components/vc/ (VCVerificationCard, WorkflowVCChain, SimpleVCTag, SimpleWorkflowVC, VCDetailsModal, VCStatusIndicator, VerifiableCredentialBadge) - Delete MCP hooks: useMCPHealth, useMCPMetrics, useMCPServers, useMCPTools - Delete pages: AuthorizationPage, CredentialsPage, DIDExplorerPage, PackagesPage, WorkflowDeckGLTestPage - Remove MCP Servers, Tools, Performance tabs from NodeDetailPage - Remove Identity & Trust and Authorization sections from navigation config - Remove deprecated routes from App.tsx router - Fix broken imports in WorkflowDetailPage, ReasonerDetailPage - Trim src/mcp/index.ts barrel to API services + types only (no component re-exports) API services (vcApi, mcpApi), types, and non-MCP hooks are preserved. TypeScript check passes with zero errors after cleanup.
… foundation system - Rewrote src/index.css with clean standard shadcn/ui theme (HSL tokens for light/dark mode) - Deleted src/styles/foundation.css (custom token system) - Rewrote tailwind.config.js to minimal shadcn-standard config (removed custom spacing, fontSize, lineHeight, transitionDuration overrides) - Replaced ~130 component files: bg-bg-*, text-text-*, border-border-*, text-nav-*, bg-nav-*, text-heading-*, text-body*, text-caption, text-label, text-display, interactive-hover, card-elevated, focus-ring, glass, gradient-* with standard shadcn equivalents - Migrated status sub-tokens (status-success-bg, status-success-light, status-success-border etc.) to opacity modifiers on base status tokens - Updated lib/theme.ts STATUS_TONES to use standard token classes - Fixed workflow-table.css status dot and node status colors to use hsl(var(--status-*)) - Zero TypeScript errors after migration
…ariants - Delete 4 JSON viewer duplicates (JsonViewer, EnhancedJsonViewer x2, AdvancedJsonViewer); all callers already use UnifiedJsonViewer - Delete 3 execution header duplicates (ExecutionHero, ExecutionHeader, EnhancedExecutionHeader); update RedesignedExecutionDetailPage to use CompactExecutionHeader - Delete 3 status indicator duplicates (ui/StatusIndicator, ui/status-indicator, reasoners/StatusIndicator); consolidate legacy StatusIndicator into UnifiedStatusIndicator module and create ReasonerStatusDot for reasoner-specific dot display - Delete RedesignedInputDataPanel and RedesignedOutputDataPanel standalone files; InputDataPanel/OutputDataPanel already export backward-compat aliases - Delete legacy Navigation/Sidebar, NavigationItem, NavigationSection (unused; SidebarNew is active) - Delete enterprise-card.tsx (no callers; card.tsx already exports cardVariants) - Delete animated-tabs.tsx; add AnimatedTabs* re-exports to tabs.tsx and update 5 callers
…dark mode default - Rewrote navigation config with 5 items: Dashboard, Runs, Agents, Playground, Settings - Built AppSidebar using shadcn Sidebar with icon-rail collapsed by default (collapsible="icon") - Built HealthStrip sticky bar showing LLM, Agent fleet, and Queue status placeholders - Built AppLayout using SidebarProvider/SidebarInset/Outlet pattern with breadcrumb header - Updated App.tsx to use AppLayout as layout route wrapper, removing old SidebarNew/TopNavigation - Added placeholder routes for /runs, /playground and their detail pages - Set defaultTheme="dark" for dark-first UI - All existing pages (Dashboard, Executions, Workflows, Nodes, Reasoners) preserved under new layout
…ents, health - Install @tanstack/react-query v5 - Create src/lib/query-client.ts with 30s stale time, 5min GC, retry=1 - Wrap App with QueryClientProvider - Add src/hooks/queries/ with useRuns, useRunDAG, useStepDetail, useAgents, useLLMHealth, useQueueStatus, useCancelExecution, usePauseExecution, useResumeExecution - Barrel export via src/hooks/queries/index.ts - Hooks delegate to existing service functions (workflowsApi, executionsApi, api) - Polling: agents 10s, system health 5s, active run DAGs 3s
…lows Add RunsPage component at /runs with: - Filter bar: time range, status, and debounced search - Table with columns: Run ID, Root Reasoner, Steps, Status, Duration, Started - Checkbox row selection with bulk action bar (Compare / Cancel Running) - Paginated data via useRuns hook with Load more support - Status badge using existing badge variants (destructive/default/secondary) - Duration formatting (Xs, Xm Ys, —) - Row click navigates to /runs/:runId Wire RunsPage into App.tsx replacing the placeholder at /runs.
…iew results Adds a new /playground and /playground/:reasonerId route with: - Reasoner selector grouped by agent node - Split-pane JSON input textarea and result display - Execute button with loading state (Loader2 spinner) - View as Execution link on successful run - Recent Runs table (last 5) with Load Input shortcut - Route-sync: selecting a reasoner updates the URL path
…t runs Replaces /dashboard with NewDashboardPage — a focused, operations-first view that answers "Is anything broken? What's happening now?" rather than displaying metrics charts. The legacy enhanced dashboard is preserved at /dashboard/legacy. Key sections: - Issues banner (conditional): surfaces unhealthy LLM endpoints and queue-saturated agents via useLLMHealth / useQueueStatus polling - Recent Runs table: last 10 runs with reasoner, step count, status badge, duration, and relative start time; click navigates to detail - System Overview: 4 stat cards (Total Runs Today, Success Rate, Agents Online, Avg Run Time) backed by dashboardService + TanStack Query with auto-refresh
…ner list Replaces the /agents placeholder with a fully functional page showing each registered agent node as a collapsible Card. Each card displays status badge with live dot, last heartbeat, reasoner/skill count, health score, and an inline reasoner list fetched lazily from GET /nodes/:id/details. Supports Restart and Config actions. Auto- refreshes every 10 s via useAgents polling.
…tity, About Adds NewSettingsPage with four tabs: - General: placeholder for future system config - Observability: full webhook config (migrated from ObservabilityWebhookSettingsPage) with live forwarder status and DLQ management - Identity: DID system status, server DID display, export credentials - About: version, server URL, storage mode Updates App.tsx to route /settings to NewSettingsPage and redirect /settings/observability-webhook to /settings.
Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
Adds RunDetailPage (/runs/:runId) as the primary execution inspection screen, replacing the placeholder. Features a split-panel layout with a proportional-bar execution trace tree on the left and collapsible Input/Output/Notes step detail on the right. Single-step runs skip the trace and show step detail directly. Includes smart polling for active runs and a Trace/Graph toggle (graph view placeholder). New files: - src/pages/RunDetailPage.tsx — main page, wires useRunDAG + state - src/components/RunTrace.tsx — recursive trace tree with duration bars - src/components/StepDetail.tsx — step I/O panel with collapsible sections Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
- Replace 4 stat cards with a horizontal stats strip above the table - Fix duration formatter to handle hours and days (e.g. "31d 6h", "5h 23m") - Compact table rows: TableHead h-8 px-3 text-[11px], TableCell px-3 py-1.5 - Table text reduced to text-xs for all data columns - Remove double padding — page container is now plain flex col gap-4 - Remove Separator between CardHeader and table - Tighten CardHeader to py-3 px-4 with text-sm font-medium title - Limit recent runs to 15 (up from 10) - Fix "View All" link to navigate to /runs instead of /workflows - Remove unused StatCard component and Clock/XCircle imports
…clickable headers - Cards start collapsed by default (was open): prevents 300+ item flood with 15 agents × 20+ reasoners - Entire card header row is the expand/collapse trigger (was isolated chevron button on far right) - Reasoner rows reduced to py-1 ~24px (was ~40px with tree characters) - Removed tree characters (├──), replaced with clean font-mono list - Play button always visible (was hidden on hover) with icon + label - Truncate reasoner list at 5, "Show N more" link to expand - Removed Config button and Restart text label — icon-only restart button - Removed redundant "15 TOTAL" badge from page header - Replaced space-y-* with flex flex-col gap-2 for card list - Removed Card/CardHeader/CardContent/Collapsible/Separator — plain divs for density
- TableHead height reduced from h-10 to h-8, padding px-4 → px-3, text-[11px] - TableCell padding reduced from p-4 to px-3 py-1.5 across all row cells - Table base text changed from text-sm to text-xs for dense data display - Run ID and Started cells use text-[11px], Reasoner cell uses text-xs font-medium - Steps and Duration cells use tabular-nums for numeric alignment - formatDuration now handles ms, seconds, minutes, hours, and days correctly - space-y-4 → space-y-3 and mb-4 → mb-3 for tighter page layout
…imestamps
- Rewrite AgentsPage from bordered cards to a borderless divide-y list inside a single Card
- Fix formatRelativeTime to guard against bogus/epoch timestamps (was showing '739709d ago')
- Expanded reasoner rows now render inline (bg-muted/30, pl-8, text-[11px]) instead of in a nested Card
- Remove page <h1> heading from AgentsPage — breadcrumb in AppLayout already identifies the page
- Add delayDuration={300} to HealthStrip TooltipProvider so tooltips don't appear immediately
- navigation.ts already correct (5 items, correct icons) — no change needed
- Dashboard already reads runsQuery.data?.workflows and navigates to /runs — no change needed
…ips, theme toggle - Use useSidebar() state to conditionally render logo text vs icon-only in collapsed mode, eliminating text overflow/clipping behind the icon rail - Add SidebarRail for drag-to-resize handle on desktop - Add SidebarSeparator between header and nav content for visual separation - Implement ModeToggle in SidebarFooter (sun/moon theme toggle, centered when collapsed) - Replace bg-primary/text-primary-foreground with bg-sidebar-primary/text-sidebar-primary-foreground in logo icon container to use correct semantic sidebar tokens - Use text-sidebar-foreground and text-sidebar-foreground/60 for logo text - Add tooltip="AgentField" to logo SidebarMenuButton so collapsed state shows tooltip on hover - Header bar: use border-sidebar-border and bg-sidebar/30 backdrop-blur instead of border-border
…result linking - Add cURL dropdown with sync and async variants; clipboard copy with "Copied!" feedback - Add collapsible schema section showing input_schema and output_schema when a reasoner is selected - Show status badge and duration in Result card header after execution - Replace "View as Execution" with "View Run →" linking to /runs/:runId - Add "Replay" button to re-run with same input
…observability check - AppLayout: change SidebarProvider defaultOpen from false to true so sidebar shows labels on first load (users can collapse via Cmd+B) - Settings/General: replace empty placeholder with useful content — API endpoint display with copy button and quick-start env var snippet - Settings/Identity: fix Server DID display — was incorrectly showing res.message (a status string) as the DID; now fetches the actual DID from /api/v1/did/agentfield-server and displays it with a copy button; shows "DID system not configured" when unavailable (local mode) - Settings: default tab remains "general" which is now useful content - Settings/Observability: tab already has full webhook config, status, DLQ management — no changes needed
…uttons - Dashboard: routes already correct (/runs/:runId and /runs) - Playground: View Run link already uses /runs/:runId - HealthStrip: connected to real data (useLLMHealth, useQueueStatus, useAgents) - RunsPage: added agent filter Select, functional Compare Selected and Cancel Running buttons - RunDetailPage: removed broken Trace/Graph toggle (Tabs/ViewMode were declared but unused), added Cancel Run button (useCancelExecution) for running runs and Replay button for failed/timeout runs Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
…opy, VC export - StepDetail: replace plain <pre> blocks with JsonHighlight component (regex-based coloring for keys, strings, numbers, booleans, null) - StepDetail: add copy-action row (Copy cURL, Copy Input, Copy Output) with transient check-icon feedback after clipboard write - RunDetailPage: add Export VC button in header that opens the /api/v1/did/workflow/:id/vc-chain endpoint in a new tab - RunTrace: extend formatDuration to handle hours (Xh Ym) and days (Xd Yh)
…order, status dots - Add click-to-sort on Status, Steps, Duration, and Started headers with asc/desc arrow indicators; sort state flows through useRuns to the API - Reorder columns: Status | Reasoner | Agent | Steps | Duration | Started | Run ID (status first for scannability, run ID de-emphasised at the far right) - Add Agent column showing agent_id / agent_name per row - Replace Badge with a compact StatusDot (coloured dot + short label) for denser status display in table rows - Update search placeholder to "Search runs, reasoners, agents…" to reflect multi-field search capability - Import cn from @/lib/utils for conditional class merging
Wire up the existing WorkflowDAGViewer component into the Run Detail page as a proper Graph tab alongside the Trace view. Multi-step runs show a Trace/Graph toggle in the header; single-step runs skip the toggle entirely and show step detail directly. Clicking a node in the graph panel selects the step and populates the right-hand detail panel.
…mpty state - Add copy button next to each Run ID (copies full ID to clipboard) - Combine Agent + Reasoner columns into a single "Target" column showing agent.reasoner in monospace (agent part muted, reasoner part primary) - Remove separate Agent column; new order: Status | Target | Steps | Duration | Started | Run ID - Add HoverCard on reasoner cell that lazily fetches and displays root execution input/output preview (only when root_execution_id is present) - Replace plain "No runs found" cell with a centered empty state using Play icon and context-aware helper text - TypeScript: 0 errors
…n, active sidebar - RunDetailPage: flex column layout with h-[calc(100vh-8rem)] so trace/step panels fill the viewport instead of using fixed 500px heights - Reorganized header: status badge and DID badge inline with title, subtitle shows workflow name + step count + duration - Added Replay button (navigates to playground with agent/reasoner target) - Added Copy ID button for quick clipboard access to the run ID - Replaced single Export VC button with an Export dropdown containing "Export VC Chain" and "Export Audit Log" (downloads JSON) - AppSidebar: active nav item now renders a left-edge accent bar (before:w-0.5 bg-sidebar-primary) for clear visual distinction in both light and dark mode, supplementing the existing bg-sidebar-accent fill
…roup separators
- Add sequential step numbers (1-based) on every trace row for disambiguation
- Show relative start times per step ("+0:00", "+1:23") anchored to run start
- Color-code duration bars: green=succeeded, red=failed, amber=timeout, blue/pulse=running
- Replace large status icons with compact inline status dots (size-1.5)
- Add group count badge (×N) on first node of consecutive same-reasoner runs
- Add subtle border separator when reasoner_id changes between siblings
- Reduce row height to py-1 (28px) for better visual density
- Pass runStartedAt prop from RunDetailPage down to RunTrace
Adds a CommandPalette component using shadcn Command + Dialog, registered globally via AppLayout. Cmd+K / Ctrl+K toggles the palette; items navigate to Dashboard, Runs, Agents, Playground, Settings, and filtered run views. A ⌘K hint badge is shown in the header bar on medium+ screens.
- Use constant-time comparison (crypto/subtle) for bearer token validation in Go SDK process logs endpoint - Add MaxBytesReader (10 MiB) to execution logs ingestion handler to prevent memory exhaustion from oversized payloads - Remove accidentally committed .cursor/ IDE state files - Add .cursor/ to .gitignore Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
The Python SDK logger only wrote structured logs to stdout, unlike the Go SDK which also dispatches them to POST /api/v1/executions/:id/logs. This caused the Logs tab in the UI to always show empty for Python agents. - Add post_execution_logs() to AgentFieldClient - Add _dispatch_to_cp() in logger to async-POST records to the CP - Wire the client into the logger during Agent init via set_cp_client() Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
…nline The HealthStrip filtered on health_status === "ready" | "active" but agents register with lifecycle_status: "ready" and health_status may be "inactive" or "unknown". This caused "0/5 online" while the table showed all agents as ready with green dots. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
santoshkumarradha
force-pushed
the
feat/ui-revamp-product-research
branch
from
aa3b4ed to
77511ab
Compare
…dback - Restore AgentLegend from pre-regression (commit 32081e3): compact embedded bar with agent color dots, Popover agent picker, Maximize2 expand-to-fullscreen button, and inline viewport controls (fit/zoom in/zoom out) - Fix runs table row click: remove stopPropagation on Target cell so clicking anywhere on the row navigates to run detail - Add copy feedback on run ID chip: show Check icon + green border for 2s after copying, matching the standard CopyButton pattern used elsewhere
Approval flow fixes: - Allow timeout→running state transition so approvals resolve even after the stale reaper marks an execution as timed out - Webhook handler no longer rejects approvals when execution was reaped but approval_status is still pending - Stale reaper now syncs both executions and workflow_executions tables in the same transaction to prevent split-brain status - Reaper skips executions with approval_status=pending (legitimately waiting for human input, not stale) - Approve/Reject buttons show loading state, handle errors, and invalidate queries to refresh the UI SSE connection exhaustion fix: - Disable node and reasoner SSE streams to stay within browser HTTP/1.1 per-origin connection limit (6); queries already include adaptive polling fallback - Execution events SSE handler sends initial connected event so browser EventSource detects the open immediately - Execution log SSE stream checks terminal status on startup and each heartbeat, closing gracefully instead of hanging forever - Heartbeats no longer reset the idle timer (was preventing timeout) Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
The ApproveWithContextDialog opened an empty tag selector with a disabled Approve button when the agent had no proposed_tags. Now it shows an "Approve Agent" flow explaining the agent registered without tags and lets the admin approve the registration directly. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
The Replay button navigated to the playground without passing the execution's input data, leaving the input editor empty. Now it passes input_data via React Router location state, and the playground reads it to seed the JSON editor on mount. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
The Replay button fetches execution details to get the original input_data before navigating to the playground. The playground reads replay input from location state and skips the schema-based empty seed when replay data is present. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
AbirAbbas
approved these changes
Apr 6, 2026
AbirAbbas
added a commit
that referenced
this pull request
Apr 6, 2026
All 11 conflicts resolved by taking the main branch versions, which contain the tested bug fixes from the UI revamp review: - Security: constant-time token comparison, MaxBytesReader body limit - SSE: terminal state detection, disabled node/reasoner streams (HTTP/1.1 limit) - Approval: handleApproval callback with busy state and query invalidation - Replay: async input fetch from execution details API - Health strip: lifecycle_status "ready" counted as online - Logger: structured log dispatch to control plane - Runs page: copy-to-clipboard feedback with Check icon - AgentLegend: proper prop usage instead of underscore-prefixed unused vars Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
github-merge-queue bot
pushed a commit
that referenced
this pull request
Apr 6, 2026
* feat(sdk/ts): add multimodal helpers for image, audio, file inputs/outputs * refactor(ui): migrate icons from Phosphor to Lucide Replace all @phosphor-icons/react imports with lucide-react equivalents. Rewrote icon-bridge.tsx to re-export Lucide icons under the same names used throughout the codebase, so no consumer files needed changing. Updated icon.tsx to use Lucide directly. Removed weight= props from badge.tsx, segmented-status-filter.tsx, and ReasonerCard.tsx since Lucide does not support the Phosphor weight API. * refactor(ui): remove MCP, Authorization, DID/VC, Packages pages — features being redesigned - Delete src/components/mcp/ (MCPServerList, MCPServerCard, MCPHealthIndicator, MCPServerControls, MCPToolExplorer, MCPToolTester) - Delete src/components/authorization/ (AccessRulesTab, AgentTagsTab, ApproveWithContextDialog, PolicyContextPanel, PolicyFormDialog, RevokeDialog) - Delete src/components/packages/ (AgentPackageCard, AgentPackageList) - Delete src/components/did/ (DIDIdentityCard, DIDDisplay, DIDStatusBadge, DIDInfoModal, DIDIndicator) - Delete src/components/vc/ (VCVerificationCard, WorkflowVCChain, SimpleVCTag, SimpleWorkflowVC, VCDetailsModal, VCStatusIndicator, VerifiableCredentialBadge) - Delete MCP hooks: useMCPHealth, useMCPMetrics, useMCPServers, useMCPTools - Delete pages: AuthorizationPage, CredentialsPage, DIDExplorerPage, PackagesPage, WorkflowDeckGLTestPage - Remove MCP Servers, Tools, Performance tabs from NodeDetailPage - Remove Identity & Trust and Authorization sections from navigation config - Remove deprecated routes from App.tsx router - Fix broken imports in WorkflowDetailPage, ReasonerDetailPage - Trim src/mcp/index.ts barrel to API services + types only (no component re-exports) API services (vcApi, mcpApi), types, and non-MCP hooks are preserved. TypeScript check passes with zero errors after cleanup. * refactor(ui): migrate to standard shadcn design tokens, remove custom foundation system - Rewrote src/index.css with clean standard shadcn/ui theme (HSL tokens for light/dark mode) - Deleted src/styles/foundation.css (custom token system) - Rewrote tailwind.config.js to minimal shadcn-standard config (removed custom spacing, fontSize, lineHeight, transitionDuration overrides) - Replaced ~130 component files: bg-bg-*, text-text-*, border-border-*, text-nav-*, bg-nav-*, text-heading-*, text-body*, text-caption, text-label, text-display, interactive-hover, card-elevated, focus-ring, glass, gradient-* with standard shadcn equivalents - Migrated status sub-tokens (status-success-bg, status-success-light, status-success-border etc.) to opacity modifiers on base status tokens - Updated lib/theme.ts STATUS_TONES to use standard token classes - Fixed workflow-table.css status dot and node status colors to use hsl(var(--status-*)) - Zero TypeScript errors after migration * refactor(ui): remove 15 duplicate components, consolidate to single variants - Delete 4 JSON viewer duplicates (JsonViewer, EnhancedJsonViewer x2, AdvancedJsonViewer); all callers already use UnifiedJsonViewer - Delete 3 execution header duplicates (ExecutionHero, ExecutionHeader, EnhancedExecutionHeader); update RedesignedExecutionDetailPage to use CompactExecutionHeader - Delete 3 status indicator duplicates (ui/StatusIndicator, ui/status-indicator, reasoners/StatusIndicator); consolidate legacy StatusIndicator into UnifiedStatusIndicator module and create ReasonerStatusDot for reasoner-specific dot display - Delete RedesignedInputDataPanel and RedesignedOutputDataPanel standalone files; InputDataPanel/OutputDataPanel already export backward-compat aliases - Delete legacy Navigation/Sidebar, NavigationItem, NavigationSection (unused; SidebarNew is active) - Delete enterprise-card.tsx (no callers; card.tsx already exports cardVariants) - Delete animated-tabs.tsx; add AnimatedTabs* re-exports to tabs.tsx and update 5 callers * feat(ui): new app shell — minimal sidebar, health strip, 5-item nav, dark mode default - Rewrote navigation config with 5 items: Dashboard, Runs, Agents, Playground, Settings - Built AppSidebar using shadcn Sidebar with icon-rail collapsed by default (collapsible="icon") - Built HealthStrip sticky bar showing LLM, Agent fleet, and Queue status placeholders - Built AppLayout using SidebarProvider/SidebarInset/Outlet pattern with breadcrumb header - Updated App.tsx to use AppLayout as layout route wrapper, removing old SidebarNew/TopNavigation - Added placeholder routes for /runs, /playground and their detail pages - Set defaultTheme="dark" for dark-first UI - All existing pages (Dashboard, Executions, Workflows, Nodes, Reasoners) preserved under new layout * feat(ui): add TanStack Query data layer with query hooks for runs, agents, health - Install @tanstack/react-query v5 - Create src/lib/query-client.ts with 30s stale time, 5min GC, retry=1 - Wrap App with QueryClientProvider - Add src/hooks/queries/ with useRuns, useRunDAG, useStepDetail, useAgents, useLLMHealth, useQueueStatus, useCancelExecution, usePauseExecution, useResumeExecution - Barrel export via src/hooks/queries/index.ts - Hooks delegate to existing service functions (workflowsApi, executionsApi, api) - Polling: agents 10s, system health 5s, active run DAGs 3s * feat(ui): build Runs page — unified view replacing Executions + Workflows Add RunsPage component at /runs with: - Filter bar: time range, status, and debounced search - Table with columns: Run ID, Root Reasoner, Steps, Status, Duration, Started - Checkbox row selection with bulk action bar (Compare / Cancel Running) - Paginated data via useRuns hook with Load more support - Status badge using existing badge variants (destructive/default/secondary) - Duration formatting (Xs, Xm Ys, —) - Row click navigates to /runs/:runId Wire RunsPage into App.tsx replacing the placeholder at /runs. * feat(ui): build Playground page — test reasoners with custom input, view results Adds a new /playground and /playground/:reasonerId route with: - Reasoner selector grouped by agent node - Split-pane JSON input textarea and result display - Execute button with loading state (Loader2 spinner) - View as Execution link on successful run - Recent Runs table (last 5) with Load Input shortcut - Route-sync: selecting a reasoner updates the URL path * feat(ui): new operations-first dashboard with issues banner and recent runs Replaces /dashboard with NewDashboardPage — a focused, operations-first view that answers "Is anything broken? What's happening now?" rather than displaying metrics charts. The legacy enhanced dashboard is preserved at /dashboard/legacy. Key sections: - Issues banner (conditional): surfaces unhealthy LLM endpoints and queue-saturated agents via useLLMHealth / useQueueStatus polling - Recent Runs table: last 10 runs with reasoner, step count, status badge, duration, and relative start time; click navigates to detail - System Overview: 4 stat cards (Total Runs Today, Success Rate, Agents Online, Avg Run Time) backed by dashboardService + TanStack Query with auto-refresh * feat(ui): build simplified Agents page — node cards with inline reasoner list Replaces the /agents placeholder with a fully functional page showing each registered agent node as a collapsible Card. Each card displays status badge with live dot, last heartbeat, reasoner/skill count, health score, and an inline reasoner list fetched lazily from GET /nodes/:id/details. Supports Restart and Config actions. Auto- refreshes every 10 s via useAgents polling. * feat(ui): build Settings page — tabs for General, Observability, Identity, About Adds NewSettingsPage with four tabs: - General: placeholder for future system config - Observability: full webhook config (migrated from ObservabilityWebhookSettingsPage) with live forwarder status and DLQ management - Identity: DID system status, server DID display, export credentials - About: version, server URL, storage mode Updates App.tsx to route /settings to NewSettingsPage and redirect /settings/observability-webhook to /settings. * feat(ui): add URL redirects from old routes to new pages Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * feat(ui): build Run Detail page — trace view with step detail panel Adds RunDetailPage (/runs/:runId) as the primary execution inspection screen, replacing the placeholder. Features a split-panel layout with a proportional-bar execution trace tree on the left and collapsible Input/Output/Notes step detail on the right. Single-step runs skip the trace and show step detail directly. Includes smart polling for active runs and a Trace/Graph toggle (graph view placeholder). New files: - src/pages/RunDetailPage.tsx — main page, wires useRunDAG + state - src/components/RunTrace.tsx — recursive trace tree with duration bars - src/components/StepDetail.tsx — step I/O panel with collapsible sections Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * fix(ui): dashboard — compact rows, stats strip, fix duration formatting - Replace 4 stat cards with a horizontal stats strip above the table - Fix duration formatter to handle hours and days (e.g. "31d 6h", "5h 23m") - Compact table rows: TableHead h-8 px-3 text-[11px], TableCell px-3 py-1.5 - Table text reduced to text-xs for all data columns - Remove double padding — page container is now plain flex col gap-4 - Remove Separator between CardHeader and table - Tighten CardHeader to py-3 px-4 with text-sm font-medium title - Limit recent runs to 15 (up from 10) - Fix "View All" link to navigate to /runs instead of /workflows - Remove unused StatCard component and Clock/XCircle imports * fix(ui): agents page — compact collapsed cards, dense reasoner list, clickable headers - Cards start collapsed by default (was open): prevents 300+ item flood with 15 agents × 20+ reasoners - Entire card header row is the expand/collapse trigger (was isolated chevron button on far right) - Reasoner rows reduced to py-1 ~24px (was ~40px with tree characters) - Removed tree characters (├──), replaced with clean font-mono list - Play button always visible (was hidden on hover) with icon + label - Truncate reasoner list at 5, "Show N more" link to expand - Removed Config button and Restart text label — icon-only restart button - Removed redundant "15 TOTAL" badge from page header - Replaced space-y-* with flex flex-col gap-2 for card list - Removed Card/CardHeader/CardContent/Collapsible/Separator — plain divs for density * fix(ui): runs page — compact table rows, fix duration formatting - TableHead height reduced from h-10 to h-8, padding px-4 → px-3, text-[11px] - TableCell padding reduced from p-4 to px-3 py-1.5 across all row cells - Table base text changed from text-sm to text-xs for dense data display - Run ID and Started cells use text-[11px], Reasoner cell uses text-xs font-medium - Steps and Duration cells use tabular-nums for numeric alignment - formatDuration now handles ms, seconds, minutes, hours, and days correctly - space-y-4 → space-y-3 and mb-4 → mb-3 for tighter page layout * fix(ui): agents as clean list, fix sidebar, fix dashboard data, fix timestamps - Rewrite AgentsPage from bordered cards to a borderless divide-y list inside a single Card - Fix formatRelativeTime to guard against bogus/epoch timestamps (was showing '739709d ago') - Expanded reasoner rows now render inline (bg-muted/30, pl-8, text-[11px]) instead of in a nested Card - Remove page <h1> heading from AgentsPage — breadcrumb in AppLayout already identifies the page - Add delayDuration={300} to HealthStrip TooltipProvider so tooltips don't appear immediately - navigation.ts already correct (5 items, correct icons) — no change needed - Dashboard already reads runsQuery.data?.workflows and navigates to /runs — no change needed * fix(ui): sidebar — proper shadcn implementation, icon-rail with tooltips, theme toggle - Use useSidebar() state to conditionally render logo text vs icon-only in collapsed mode, eliminating text overflow/clipping behind the icon rail - Add SidebarRail for drag-to-resize handle on desktop - Add SidebarSeparator between header and nav content for visual separation - Implement ModeToggle in SidebarFooter (sun/moon theme toggle, centered when collapsed) - Replace bg-primary/text-primary-foreground with bg-sidebar-primary/text-sidebar-primary-foreground in logo icon container to use correct semantic sidebar tokens - Use text-sidebar-foreground and text-sidebar-foreground/60 for logo text - Add tooltip="AgentField" to logo SidebarMenuButton so collapsed state shows tooltip on hover - Header bar: use border-sidebar-border and bg-sidebar/30 backdrop-blur instead of border-border * feat(ui): playground — cURL copy, async cURL, schema display, better result linking - Add cURL dropdown with sync and async variants; clipboard copy with "Copied!" feedback - Add collapsible schema section showing input_schema and output_schema when a reasoner is selected - Show status badge and duration in Result card header after execution - Replace "View as Execution" with "View Run →" linking to /runs/:runId - Add "Replay" button to re-run with same input * fix(ui): sidebar default open, settings — API info, fix DID display, observability check - AppLayout: change SidebarProvider defaultOpen from false to true so sidebar shows labels on first load (users can collapse via Cmd+B) - Settings/General: replace empty placeholder with useful content — API endpoint display with copy button and quick-start env var snippet - Settings/Identity: fix Server DID display — was incorrectly showing res.message (a status string) as the DID; now fetches the actual DID from /api/v1/did/agentfield-server and displays it with a copy button; shows "DID system not configured" when unavailable (local mode) - Settings: default tab remains "general" which is now useful content - Settings/Observability: tab already has full webhook config, status, DLQ management — no changes needed * fix(ui): P0 fixes — routes, health strip data, agent filter, action buttons - Dashboard: routes already correct (/runs/:runId and /runs) - Playground: View Run link already uses /runs/:runId - HealthStrip: connected to real data (useLLMHealth, useQueueStatus, useAgents) - RunsPage: added agent filter Select, functional Compare Selected and Cancel Running buttons - RunDetailPage: removed broken Trace/Graph toggle (Tabs/ViewMode were declared but unused), added Cancel Run button (useCancelExecution) for running runs and Replay button for failed/timeout runs Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * feat(ui): step detail — JSON syntax highlighting, cURL/input/output copy, VC export - StepDetail: replace plain <pre> blocks with JsonHighlight component (regex-based coloring for keys, strings, numbers, booleans, null) - StepDetail: add copy-action row (Copy cURL, Copy Input, Copy Output) with transient check-icon feedback after clipboard write - RunDetailPage: add Export VC button in header that opens the /api/v1/did/workflow/:id/vc-chain endpoint in a new tab - RunTrace: extend formatDuration to handle hours (Xh Ym) and days (Xd Yh) * feat(ui): runs table — sortable columns, agent column, better column order, status dots - Add click-to-sort on Status, Steps, Duration, and Started headers with asc/desc arrow indicators; sort state flows through useRuns to the API - Reorder columns: Status | Reasoner | Agent | Steps | Duration | Started | Run ID (status first for scannability, run ID de-emphasised at the far right) - Add Agent column showing agent_id / agent_name per row - Replace Badge with a compact StatusDot (coloured dot + short label) for denser status display in table rows - Update search placeholder to "Search runs, reasoners, agents…" to reflect multi-field search capability - Import cn from @/lib/utils for conditional class merging * feat(ui): run detail — integrate ReactFlow DAG as graph view toggle Wire up the existing WorkflowDAGViewer component into the Run Detail page as a proper Graph tab alongside the Trace view. Multi-step runs show a Trace/Graph toggle in the header; single-step runs skip the toggle entirely and show step detail directly. Clicking a node in the graph panel selects the step and populates the right-hand detail panel. * feat(ui): runs table — ID copy, agent.reasoner format, I/O preview, empty state - Add copy button next to each Run ID (copies full ID to clipboard) - Combine Agent + Reasoner columns into a single "Target" column showing agent.reasoner in monospace (agent part muted, reasoner part primary) - Remove separate Agent column; new order: Status | Target | Steps | Duration | Started | Run ID - Add HoverCard on reasoner cell that lazily fetches and displays root execution input/output preview (only when root_execution_id is present) - Replace plain "No runs found" cell with a centered empty state using Play icon and context-aware helper text - TypeScript: 0 errors * feat(ui): run detail — full height, replay, DID badge, export dropdown, active sidebar - RunDetailPage: flex column layout with h-[calc(100vh-8rem)] so trace/step panels fill the viewport instead of using fixed 500px heights - Reorganized header: status badge and DID badge inline with title, subtitle shows workflow name + step count + duration - Added Replay button (navigates to playground with agent/reasoner target) - Added Copy ID button for quick clipboard access to the run ID - Replaced single Export VC button with an Export dropdown containing "Export VC Chain" and "Export Audit Log" (downloads JSON) - AppSidebar: active nav item now renders a left-edge accent bar (before:w-0.5 bg-sidebar-primary) for clear visual distinction in both light and dark mode, supplementing the existing bg-sidebar-accent fill * feat(ui): trace view — step numbers, relative times, status colors, group separators - Add sequential step numbers (1-based) on every trace row for disambiguation - Show relative start times per step ("+0:00", "+1:23") anchored to run start - Color-code duration bars: green=succeeded, red=failed, amber=timeout, blue/pulse=running - Replace large status icons with compact inline status dots (size-1.5) - Add group count badge (×N) on first node of consecutive same-reasoner runs - Add subtle border separator when reasoner_id changes between siblings - Reduce row height to py-1 (28px) for better visual density - Pass runStartedAt prop from RunDetailPage down to RunTrace * feat(ui): command palette — Cmd+K for navigation and quick actions Adds a CommandPalette component using shadcn Command + Dialog, registered globally via AppLayout. Cmd+K / Ctrl+K toggles the palette; items navigate to Dashboard, Runs, Agents, Playground, Settings, and filtered run views. A ⌘K hint badge is shown in the header bar on medium+ screens. * feat(ui): run comparison page — side-by-side step diff with divergence highlighting Adds /runs/compare?a=RUN_ID_1&b=RUN_ID_2 route accessible from the Runs page when exactly 2 runs are selected via "Compare Selected". The page shows dual summary cards (status, step count, failures, duration delta), a step-by-step alignment table with same/diverged/extra annotations, and a clickable output-diff panel for diverged rows. Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * feat(ui): virtual scrolling for trace, HITL approval UI in step detail - Flatten the recursive RunTrace tree and virtualize with @tanstack/react-virtual for performant rendering of 200+ step traces (overscan=20, estimateSize=28px) - Extract TraceRow as a standalone component; preserve all visual logic (step numbers, depth indentation, connector glyphs, status dots, bars, group-count badges, relative-start times, group separators) - Add HITL approval card in StepDetail: shows when execution.status === "waiting" or approval_request_id is present; displays approval_status badge and approval_requested_at timestamp; renders Approve/Reject buttons when approval_status === "pending", posting to /api/v1/webhooks/approval-response Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * feat(ui): SSE live updates, connection indicator, webhook delivery status in step detail - Add useSSEQuerySync hook that subscribes to execution and node SSE channels and invalidates TanStack Query caches on every event, so Runs, Run DAG, and Step Detail pages auto-refresh without polling - Mount useSSEQuerySync once in AppLayout for app-wide coverage - Add SSE connection status indicator (Live / Reconnecting / Disconnected) to HealthStrip so users know whether real-time updates are active - Add Webhook Delivery collapsible section to StepDetail showing per-event HTTP status, delivery time, and a Retry button for failed deliveries; only rendered when webhook_registered or webhook_events are present Co-Authored-By: Claude Sonnet 4.6 <[email protected]> * fix(ui): remove double scroll in trace, add I/O diff to comparison, sticky header - RunDetailPage: remove ScrollArea wrapper around RunTrace; the virtualizer already owns its scroll container (h-full overflow-auto), so the outer ScrollArea caused scroll-inside-scroll - ComparisonPage: replace status-only OutputDiff with StepDiff that fetches both executions via useStepDetail and renders input/output JSON side-by-side with error highlighting; all rows are now clickable (not just diverged) - ComparisonPage: make summary-cards section sticky (top-0 z-10) so it stays visible while scrolling the step comparison table * fix(ui): graph view — edges visible, node clicks populate right panel instead of sidebar - Destructure onExecutionClick in WorkflowDAGViewerInner (was defined in props interface but never used inside the component) - handleNodeClick and handleDeckNodeClick now call onExecutionClick when provided, falling back to the internal NodeDetailSidebar only when no parent handler is registered (preserves legacy WorkflowDetailPage usage) - Suppress internal NodeDetailSidebar render when onExecutionClick is set - FloatingEdge: remove early return null when useInternalNode yields undefined; fall back to React Flow's default handle coordinates (sourceX/sourceY/targetX/targetY) so edges render before the first measure cycle completes - Fix duplicate SVG marker ids by scoping id to the edge id, preventing cross-edge marker conflicts * fix(ui): graph edges invisible — wrap HSL vars in hsl() for SVG stroke attributes * web(ui): revamp layout, filters, and health strip - Update AppLayout, AppSidebar, and global styles - Add filter combobox components and popover - Refresh Agents and Runs pages with new filtering UX - Adjust HealthStrip; update client dependencies * feat(web): UI revamp — components, pages, logos, and shared UI utilities - Refresh control-plane web client pages and execution/reasoner components - Add logo assets, pagination, JSON highlight, endpoint icons, reasoner combobox - Add reasonerCompareExtract util; update Tailwind and global styles - Track Cursor continual-learning hook state artifact * feat(ui): inset shell, unified theme tokens, runs workflow updates - Sidebar inset variant with rounded rail; SidebarInset overflow/min-h-0; single main landmark - Dark neutrals use shared --shell-h/--shell-s; accent isolated to --brand-* (logo, rings, sidebar-primary) - Run detail trace/step UI, executions API/handler, scroll-area, ui-revamp docs * fix(web-ui): run detail workflow graph renders with valid React Flow bounds - Replace embedded WorkflowGraphViewport absolute inset-0 with flex layout so the pane gets non-zero height in the document flow (fixes xyflow error #4). - Give flow containers explicit minHeight/width/flex and pass width/height styles into ReactFlow and VirtualizedDAG. - Defer initial fitView/setViewport until after nodes commit (double rAF). - Reset viewport/layout only on workflow id change, not first mount; RunDetail passes stable workflowId, key=runId, and graph column flex/min-height. * feat(ui): runs/step detail, DAG polish, execution APIs, provenance card - Backend: UI executions and workflow DAG handler updates. - Web: RunsPage, StepDetail, StepProvenanceCard; WorkflowNode and AgentLegend tweaks. - Client types and executions/vc API services aligned with server payloads. - Cursor continual-learning hook state snapshot. * feat(web): workflow DAG controls, dashboard/settings polish, runs query updates - Add WorkflowGraphControls and integrate with WorkflowDAG - Update AppLayout, AgentLegend, DeckGLView, VirtualizedDAG - Refresh NewDashboardPage, NewSettingsPage, alert component - Adjust useRuns hook; sync continual-learning hook state * feat(web): playground and reasoner combobox updates - Update PlaygroundPage - Refine reasoner-node-combobox UI - Sync continual-learning hook state * feat: access management UI, verify provenance, governance hooks - Add AccessManagementPage and authorization components (policies, tags, dialogs) - Add VerifyProvenancePage and vcApi/did type updates; af verify-provenance CLI + tests - Wire DID handlers, UI DID routes, API catalog, and server routes - Governance queries/probe/utils; navigation, App, CommandPalette, Agents/Run detail polish - Sync continual-learning hook state * fix(web): app shell, navigation, run detail, verify provenance polish - Tweak AppLayout and CommandPalette - Update navigation config - Refine RunDetailPage and VerifyProvenancePage - Sync continual-learning hook state * feat(web): dashboard workload components and Figma/shadcn audit doc - Add DashboardActiveWorkload, DashboardRunOutcomeStrip, dashboardRunUtils - Refactor NewDashboardPage; remove NewDashboardPage.tail.tsx - Update client package.json and pnpm-lock.yaml - Add docs/FIGMA_SHADCN_AUDIT.md - Sync continual-learning hook state * style(web): broad UI polish across shell, tables, DAG, and design tokens - Refine AppLayout, sidebars, navigation, and global index.css / Tailwind config - Update executions tables, health strip, run trace, step detail, provenance card - Polish WorkflowDAG components and workflow dashboard panels - Refresh shadcn-style UI primitives (badge, button, drawer, sidebar, etc.) - Add copy-identifier-chip and entity-tag components - Touch dashboard, authorization, and key pages (agents, runs, playground, etc.) - Extend FIGMA_SHADCN_AUDIT.md; sync continual-learning hook state * chore: remove internal UI planning docs and stray tooling files Drop docs/ui-revamp and embedded client audit notes from the branch; surface product direction in the PR instead. Remove accidentally tracked .agit index and Claude worktree pointer; add ignore rules. * fix: address review findings across Go handlers, TS SDK, and web UI Multi-pass code review surfaced 60 issues across security, code quality, performance, API design, frontend, and test coverage. This commit fixes the critical, high, and medium severity items: - Go: fix ValidComponents count, immutable mergeDIDBundle, empty-DID guard, remove debug fmt.Printf, return 422/413 on verify-audit endpoints - TS SDK: fix Audio.fromFile format detection (in→includes), imageFromBuffer default mimeType, remove unnecessary async on fromUrl methods - Web: add top-level ErrorBoundary, limit auto-reset to one attempt, DEV-gate SSE console.log, opt-in events accumulation, extract SortableHead component, hoist Intl.RelativeTimeFormat, guard bulk cancel async handler, add keyboard a11y to table rows, convert sidebar nav to Link elements, wrap DAG in ErrorBoundary, fix HealthStrip false-healthy default * fix: simplify review fixes — stale state, import style, destructuring, parallel cancel - ErrorBoundary: use functional setState in auto-reset timer to avoid stale closure; also clear errorInfo/errorId on auto-reset - RunDetailPage: use @/ alias for ErrorBoundary import (consistency) - useSSE: move trackEvents into existing destructuring block - RunsPage: parallelize bulk cancel with Promise.all instead of sequential loop * refactor: extract shared handler, reuse SortableHeaderCell, consolidate formatRelativeTime - Go: extract HandleVerifyAuditBundle into handlers/verify_audit.go, both DIDHandlers.VerifyAuditBundle and DIDHandler.VerifyAuditBundleHandler now delegate to the shared function (removes 30+ lines of duplication) - Web: replace local SortableHead in RunsPage with shared SortableHeaderCell from CompactTable.tsx (reuses existing exported component) - Web: add formatCompactRelativeTime to utils/dateFormat.ts and replace 5 local copies across AgentsPage, NodesPage, CompactReasonersStats, CompactWorkflowSummary, and ExecutionTimeline * fix: move ErrorBoundary inside Router to preserve routing context * style(web): redesign sign-in page with shadcn components and app branding - AuthGuard: replace raw HTML inputs/button with shadcn Card, Input, Button, Alert; add app logo (light/dark), icon-prefixed fields (KeyRound, ShieldCheck), Loader2 spinner, proper labels for a11y, and friendlier error messages - AuthContext: replace plain "Loading..." text with a styled spinner and "Connecting…" label matching the app's design tokens * fix: resolve all tsc -b strict build errors for CI - App.tsx: remove 8 unused page imports - DeckGLView.tsx: narrow zoom type (number | [number, number]) before arithmetic - VirtualizedDAG.tsx: cast nodeTypes/edgeTypes to satisfy xyflow strict types - WorkflowDAG/index.tsx: cast DeckGL node click handler and component types - AgentLegend.tsx: restore main's version, export layout types for callers - AccessRulesTab.tsx, PolicyFormDialog.tsx: remove invalid weight prop from lucide icons - reasoner-node-combobox.tsx: add boolean coercion for string | boolean - EnhancedDashboardPage.tsx: change "warning" to valid "degraded" badge variant - ExecutionDetailPage.tsx: create ExecutionHeader component, remove unused import - NodeDetailPage.tsx: restore main's version with latestEvent bindings - reasonerCompareExtract.ts: fix number | boolean comparison with explicit ternary - Add stub modules for DIDInfoModal, mcp/index, animated-tabs * security: prevent SSRF in DID resolution and custom resolver endpoints - Add safeHTTPClient with DialContext that resolves DNS and blocks private/loopback/link-local IPs (10/8, 172.16/12, 192.168/16, 127/8, 169.254/16, ::1, fc00::/7, fe80::/10) - Add validateExternalURL requiring HTTPS scheme - Apply to resolveWebDID and resolveFromCustom (CodeQL critical findings) - Add io.LimitReader (2 MiB) on all external response bodies to prevent memory exhaustion from malicious DID document hosts - Set 10s client timeout and 5s dial timeout * chore: stacked branch for UI live-update follow-up work Establishes a PR stack on top of #330. No functional changes yet. * feat(web): unified SSE sync and adaptive polling for live UI - SSESyncProvider runs executions, nodes, and reasoners SSE once; invalidates runs, run-dag, step-detail, agents, reasoners, and dashboard-summary - HealthStrip uses shared connection state; Refresh resyncs when streams are down - Faster TanStack refetch intervals when SSE is disconnected (runs, agents, LLM/queue, dashboard summary); explicit run list intervals tighten too - useRunDAG/useStepDetail poll active steps more often, especially without SSE - AllReasonersPage uses React Query with SSE-driven invalidation and 6s fallback * fix(web): match polling and badges to the SSE channel that feeds each view - useRuns, run DAG/step detail, dashboard summary, LLM/queue: gate on execConnected - useAgents: gate on nodeConnected - All Reasoners: fallback poll when neither node nor reasoner stream is up - HealthStrip Live/Refresh tied to execution stream (runs/steps) * agit sync * agit sync * fix(web): tighten live update gating * feat: agent node process logs (Python NDJSON, CP proxy, UI panel) - Document NDJSON v1 contract and UI proxy path in docs/api/AGENT_NODE_LOGS.md - Python SDK: ring buffer, stdio tee, GET /agentfield/v1/logs with internal bearer - Control plane: proxy GET /api/ui/v1/nodes/:nodeId/logs, node-log-proxy settings GET/PUT - Web UI: NodeProcessLogsPanel on Agents expanded row and Node detail Logs tab; Settings tab for proxy limits using existing shadcn cards and patterns * docs: document node log proxy and agent log env vars in .env.example * fix: align .env.example log vars with Python node_logs.py * agit sync * agit sync * chore: log-demo compose, UI proxy functional test, example heartbeats - Makefile targets log-demo-up / log-demo-down - docker-compose.log-demo.yml + Node log-demo agent Dockerfile - test_ui_proxies_node_process_logs exercises GET /api/ui/v1/nodes/:id/logs - Python + Go demo agents emit periodic stdout/stderr for UI verification - Trim unused notify from Go process log ring; drop unused TS import * fix(docker): wait for CP health before log-demo agents start Add wait-control-plane (curl with retries) and depends_on completed_successfully so Python/Go/Node demo agents do not register before the API is ready. Set restart: unless-stopped on demo agents. Document behavior in LOG_DEMO.md. * feat(scripts): host log-demo stack when Docker is unavailable Add run-log-demo-native.sh / stop-log-demo-native.sh with writable paths (env overrides for SQLite, Bolt, DID keystore under /tmp/agentfield-log-demo). Makefile targets log-demo-native-up/down. Document in LOG_DEMO.md. Fixes local runs using agentfield-test.yaml which points storage at /data. * feat(ui,sdks): process log filters, structured rows, NDJSON level/source - AgentsPage: full-width tab bar aligned with node detail (icons, muted track). - NodeProcessLogsPanel: stream segmented control (All/Stdout/Stderr with counts), text search across line/seq/level/source, rows show local time + date when not today, seq, stream/level badges, optional source line; stderr row tint. - NodeLogEntry type: optional level, source. - Python/Go/TypeScript SDKs: emit level (info/error/log) and source=process for stdio capture; document in AGENT_NODE_LOGS.md. * fix: tighten audit verification and multimodal IO * feat(ui): agents page heading, nav labels, compact responsive logs - Page h1 Agent nodes & logs; sidebar/command palette use Agent nodes - Agents row: terminal shortcut + controlled tabs; breadcrumb route name - NodeProcessLogsPanel: responsive header toolbars, compact log grid, hide redundant level badges, optional native demo scripts unchanged * fix(sdk/python): drop unused Generator import (ruff F401) Unblocks Python SDK CI lint-and-test job. * fix(sdk/ts): qualify Express Response in flush cast (TS2352) Bare Response resolved to DOM Fetch Response; use import('express').Response. * fix(ci): harden DID fetches and stabilize node log proxy test * agit sync * fix(web): avoid format strings in node action logs * fix(cli): keep VC verification offline-only * fix(tests): write node log marker into process ring * docs: add execution observability RFC * agit sync * agit sync * agit sync * agit sync * agit sync * agit sync * agit sync * feat: add execution logging transport * feat(py): add structured execution logging * agit sync * feat(web): unify execution observability panel * Remove local plandb database from repo * fix demo execution observability flow * refactor run detail into execution and logs tabs * refine execution logs density * agit sync * polish execution logs panel header * refine raw node log console * polish process log filter toolbar * standardize observability spacing primitives * agit sync * test execution observability in functional harness * agit sync * test(go-sdk): add tests for types/status, types/types, types/discovery, did/types, agent/harness Adds comprehensive test coverage for previously untested Go SDK packages: - types/status_test.go: NormalizeStatus (canonical pass-through, all aliases, case-insensitive, whitespace trimming), IsTerminalStatus, IsActiveStatus, mutual-exclusivity invariant - types/types_test.go: JSON round-trip for all structs (NodeRegistrationRequest/Response, WorkflowExecutionEvent, ActionAckRequest, LeaseResponse, ShutdownRequest, NodeStatusUpdate), omitempty field validation, constant value assertions - types/discovery_test.go: JSON round-trip for DiscoveryResponse, CompactDiscoveryResponse, all capability types; optional field omission - did/types_test.go: JSON round-trip for DIDIdentity, DIDIdentityPackage, RegistrationRequest/Response, ExecutionContext, VCGenerationRequest, ExecutionVC, WorkflowVCChain; omitempty validation - agent/harness_test.go: HarnessRunner lazy initialization, singleton behavior, HarnessConfig field mapping, concurrent access safety, error propagation without provider, per-call option override * test: comprehensive test coverage audit and implementation - Add TEST_COVERAGE_AUDIT.md with full audit of all 645 source files - Control plane: memory handler tests (30 tests), nodes_rest handler tests - Python SDK: verification, node_logs, mcp_manager, mcp_stdio_bridge, serverless tests (143 tests) - TypeScript SDK: ExecutionContext, MemoryClient, WorkflowReporter, AgentRouter, MCPClient tests (110 tests) - All tests passing across Go, Python, and TypeScript * fix(tests): address quality audit — eliminate false greens and strengthen assertions - verification.py: add @pytest.mark.asyncio to async tests (were silently passing as no-ops) - mcp_manager: replace vacuous `or True` assertion with real post-condition check - memory_test: add AssertNotCalled for PublishMemoryChange when StoreEvent fails - memory_test: add delete handler event verification (action, key, previous_data) - agent_serverless: add real-registration tests (not just patch.object) - agent_serverless: verify async_mode precondition was True before asserting False - nodes_rest_test: add normalizePhase distinctness invariant test - nodes_rest_test: document that handler behavior tests require StorageProvider mock * test: add invariant and property-based tests for AI code safety Add test_invariants.py with 15 tests covering: - Status normalization: idempotency, fixed-point, disjointness, coverage - Execution context: cleanup after success/failure, no state leaking - Memory scope independence - Serialization roundtrip preservation - Discovery response schema stability - ProcessLogRing sequence monotonicity (including after eviction) - Error response structure consistency - Property-based fuzzing with hypothesis (when available) * test(ts-sdk): add behavioral invariant tests for rate limiter, DID auth, memory, agent, execution context Adds 68 invariant tests across 5 new files verifying structural properties that must always hold: circuit breaker state machine transitions, consecutive failure counter monotonicity, nonce uniqueness and timestamp ordering in DID signing, scope-to-header mapping stability in MemoryClient, reasoner/skill namespace independence in Agent, and AsyncLocalStorage scope isolation in ExecutionContext (including concurrent execution safety). * test(control-plane): add behavioral invariant tests for event bus, UoW, state machine, memory Adds 35 property-based invariant tests across four packages designed to catch regressions in AI-generated code changes by verifying structural contracts rather than specific input→output pairs. - events: ordering, silent drop on full buffer, subscriber isolation, concurrent publish safety (-race), unsubscribe channel closure, no retroactive delivery - storage/unitofwork: registration ordering, IsActive lifecycle, idempotent rollback, HasChanges correctness, register-ignored-when-inactive, ChangeType attribution - storage/execution_state: terminal irreversibility, full reachability BFS from "unknown", determinism, alias normalization equivalence, known valid/invalid transition regression guards - handlers/memory: Set-Get roundtrip, scope isolation, Delete removes, event action assertions (set/delete), hierarchical scope search order (workflow→session→actor→global) * agit sync * agit sync * test(python-sdk): add behavioral invariant tests for DID auth, memory, connection, execution state, agent lifecycle 71 invariant tests covering: - DID auth: nonce uniqueness, timestamp freshness, signature determinism/sensitivity, header completeness - Memory: scope hierarchy order, scope independence, key namespacing, null safety - Connection manager: state machine transitions, shutdown idempotency - Execution state: normalization idempotency, terminal detection, serialization roundtrip, metrics monotonicity - Agent lifecycle: registration persistence, double-registration replacement, discovery schema stability, node_id immutability * fix(tests): resolve CI lint failures — remove unused imports, fix f-strings Ruff auto-fixed 48 lint issues across all Python test files: - Remove unused imports (typing.Any, typing.Dict, asyncio, etc.) - Remove f-string prefix on strings without placeholders - Remove unused local variables - Add noqa for intentional import-for-availability check * test(go-sdk): add behavioral invariant tests for client, agent, status, harness Invariant tests covering: - Client: base URL normalization, auth header mutual exclusivity, error structure, timeout propagation - Agent: registration idempotency, config immutability, discovery schema stability NOTE: concurrent registration test skipped — found real race in RegisterReasoner (agent.go:546) - Status: normalization is a projection, terminal/active partition, alias→canonical consistency, case insensitivity - Harness: retry bounded by MaxRetries, schema repair idempotency (known limitation: multiple trailing commas) * fix(tests): add event loop fixture for Python 3.8/3.9 compatibility Agent() constructor and handle_serverless() use asyncio internally. On Python 3.8/3.9, asyncio.get_event_loop() raises RuntimeError in non-async contexts without an explicitly created loop. Add autouse fixture that ensures a loop exists. * fix(tests): remove MCP test files (unused feature), restore Go SDK source files - Remove test_mcp_manager.py, test_mcp_stdio_bridge.py, mcp_client_expanded.test.ts (MCP is not a shipped feature — no need for test coverage) - Re-skip Go SDK concurrent registration test (production fix tracked separately) - Re-skip schema repair idempotency for multiple commas (fix tracked separately) - Restore original agent.go, cli.go, schema.go (production fixes go in separate PR) * test(ts-sdk,web-ui): add skill/reasoner registry tests and Web UI service layer tests - sdk/typescript/tests/skill_registry.test.ts: 11 tests covering register, get, all(), duplicate overwrite, and includeRouter integration with AgentRouter - sdk/typescript/tests/reasoner_registry.test.ts: 13 tests covering the same surface on ReasonerRegistry, including isolation (skills not imported) - control-plane/web/client/src/test/services/api.test.ts: 15 tests for the base API client — key/token management, X-API-Key header injection, error parsing, timeout handling, and parseNodeLogsNDJSON - control-plane/web/client/src/test/services/dashboardService.test.ts: 10 tests for getDashboardSummary, getDashboardSummaryWithRetry, and getEnhancedDashboardSummary including query-string parameter forwarding - control-plane/web/client/src/test/services/executionsApi.test.ts: 19 tests for list/get/filter/pagination, field normalisation, alternative field-name fallbacks, cancelExecution, and getExecutionStats All 68 new tests pass (24 SDK + 44 Web UI). * test(storage): add integration tests for local.go — agent CRUD, executions, workflows, memory Covers the four highest-impact categories of LocalStorage operations: - Agent CRUD: register/get roundtrip, upsert on re-register, list, delete-version, not-found - AgentExecution records: store/get, query by agent ID and status, pagination - WorkflowExecution: store/get, query by workflow ID, state-machine-aware status update - Memory (BoltDB): set/get for all scopes, scope isolation, delete, list, overwrite, scope-ID isolation - Invariants: empty DB returns non-nil empty slices, stored timestamps are UTC All tests run against a real SQLite + BoltDB backed by t.TempDir(). * test(control-plane): add handler and service tests — agentic status, config storage, memory ACL, tag normalization * fix(storage): graceful LIKE fallback when SQLite FTS5 module is unavailable QueryWorkflowExecutions now checks hasFTS5 flag and falls back to LIKE-based search across key columns instead of failing with "no such table: workflow_executions_fts". * test: add 270 tests for critical coverage gaps — types, agent_server, connector handlers, MCP, ExecutionStatus Python SDK: - test_types.py: 99 tests covering all Pydantic/dataclass models, enums, serialization - test_agent_server.py: 54 tests covering routes, health, MCP ops, approval, utilities Go Control Plane: - connector/handlers_test.go: 48 tests covering all HTTP handlers, capability gating, errors TypeScript SDK: - execution_status.test.ts: 22 tests for status normalization, terminal/active classification - mcp_client.test.ts: 27 tests for health checks, tool listing, tool execution - mcp_registry.test.ts: 20 tests for client registry, tool registrar, namespace handling * fix(ci): remove unused imports — os in test_agent_server.py, beforeEach/makeFetchMock in api.test.ts * agit sync * fix(tests): align tests with main branch changes, remove .cursor files - Update execution state invariant tests: timeout is no longer fully terminal — allows transitions to running/cancelled (stale reaper fix) - Add post_execution_logs mock to MockClient (new SDK method) - Remove opencode_server test assertions (feature removed in PR #326) - Remove tracked .cursor/ IDE state files Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]> --------- Co-authored-by: Claude Sonnet 4.6 <[email protected]> Co-authored-by: Abir Abbas <[email protected]>
6 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Refreshes the embedded control plane web UI into an operations-first control surface and now also folds in the full live-update stack plus agent node process logs.
Final integrated branch tip: 5236e69
Included scope
This PR now combines three previously separate lines of work into one final integration PR:
Product direction
Review and integration notes
Issue linkage
Fixes #324.
Validation status
GitHub CI is the source of truth for this final integrated branch and is running on PR #330 after commit 5236e69.
Local verification on the integrated branch included:
npm exec -- tsc --noEmitgo test ./internal/handlers/... ./internal/storage/... ./internal/server/... -count=1tests/test_go_sdk_cli.py -k observabilityPrevious review fixes still included