Skip to content

PAI_DASHBOARD.md

Latest commit

 

History

History
38 lines (25 loc) · 3.26 KB

PAI_DASHBOARD.md

File metadata and controls

38 lines (25 loc) · 3.26 KB

PAI Control Dashboard Functionality Matrix

The PAI Control Dashboard, orchestrating the ~/.claude/MEMORY system via PMServer.ts, exposes an extensive GUI on http://localhost:8888. It acts as the primary cockpit for tracking cross-module tasks, LLM context windows, and structural awareness.

1. Global Navigation & Action Bar

The persistent header exposes immediate health metrics and global CRUD endpoints:

  • Metrics Ribbon: 7 Missions (6 Active), 19 Projects, 831 Tasks (397 Done, 0 Overdue, 0 Blocked).
  • Action Buttons:
    • + Mission: Deploys a modal interface for declaring a high-level operational objective.
    • + Project: Deploys a modal interface for instantiating a sub-goal tied to a specific mission.
  • 14 Core Subsystems: Access to Analytics, Awareness, Blockers, Board, Calendar, Email, Data, Dispatch, Git, Integration, Interview, Network, Projects, and Timeline.

2. Structural & Tracking Views

  • Analytics/Data: Granular tabular interfaces for exploring Active/Planning/Completed models. Includes full metric aggregation (completion percentage, linked task ratio) and sorting tools.
  • Board (Kanban): A global Kanban visualization managing cross-project state pipelines: Active, Planning, In Progress, Blocked, Paused, Completed.
  • Timeline (Gantt): A global rendering of all active projects across a multi-month calendar, visualizing bottlenecks and multi-step duration logic.
  • Network Maps: Force-directed 2D maps demonstrating the parent-child node graph (Mission → Project → Task). Offers semantic zooming and tag-based visual clustering.

3. Agents & Dispatch Execution

  • Dispatch Mode: Exposes API wrappers to delegate queued workflow processes to automated AI subroutines. Summarizes repo states, scopes future objectives, performs risk assessments, and auto-executes 'Next Enactable Step' strategies directly from the interface.
  • Interview Mode: Allows establishing guided, persistent multi-turn conversational agents dedicated to specific role-playing (e.g., scoping product features, designing architectural diagrams based on active projects).

4. Integration & Operations

  • Git Synchronization: Complete GUI over raw git hooks. Provides push/pull/sync buttons, connection status monitoring, and repo linking.
  • Email & Communications: Bridges tasks with outbound agent actions via AgentMail/Gmail templates. Supports configuring Ollama instances for offline parsing.
  • Data Portability: Lossless conversion and syncing. Exports operational state to JSON/CSV/Markdown. Supports state-injection via JSON data imports.

Technical Considerations

The GUI utilizes a hybrid approach for real-time responsiveness:

  • WebSocket Push (v1.1.8+): The Python WebsiteServer (:8787) establishes a stable WebSocket connection (:8890), pushing dashboard health metrics and PAI awareness state in real-time. This replaced the legacy 15-second REST polling introduced in earlier versions, significantly reducing latency and overhead.
  • REST Endpoints: Direct action data-mutating endpoints rely on the local /api REST layer, acting as a gateway to the primary PMServer.ts. Minor exceptions exist (e.g. /api/calendar/events 500 block during headless operation), but the foundation remains unconditionally resilient.