← Back to Dashboard
Contents
Overview
Mission Control is a dashboard that provides visibility into an AI agent swarm's activities, performance metrics, and task pipeline. It serves as the central nervous system for coordinating work between human operators and AI agents.
Core Concepts
Agents - AI workers with specialized roles (Developer, Wordsmith, Scout, etc.)
P&L (Profit & Loss) - A scoring system that tracks agent performance based on task outcomes and feedback sentiment
Task Pipeline - Kanban-style workflow moving tasks from Inbox to Done
Journal Sentiment - Automated scoring based on feedback patterns in conversation logs
Agent P&L Scoreboard
Hierarchical view of agent performance with expandable sections showing credits, debits, and balance for each agent and category.
MISSION CONTROL P&L
═══════════════════════════════════════════════════════════════
Credits Debits Balance
───────────────────────────────────────────────────────────────
▼ CLAUDE CODE +35 -150 -115
───────────────────────────────────────────────────────────────
▼ OPENCLAW +240 -175 +65
┌─ Growth ─────────────────────────────────────────────────
│ Entrepreneur — -75 -75
│ Scout +35 — +35
│ Subtotal +35 -75 -40
├─ Execution ──────────────────────────────────────────────
│ Developer +50 — +50
│ PM +15 — +15
│ Systems — — 0
│ Subtotal +65 — +65
├─ Content ────────────────────────────────────────────────
│ Wordsmith +105 — +105
│ Subtotal +105 — +105
└─ Operations ─────────────────────────────────────────────
Coach, Tracker, Gamemaster
═══════════════════════════════════════════════════════════════
TOTAL +275 -325 -50
Agent Categories
Category
Agents
Purpose
Growth
Entrepreneur, Scout
Business development, revenue generation
Execution
Developer, PM, Systems
Building, shipping, infrastructure
Content
Wordsmith
Writing, marketing, social media
Operations
Coach, Tracker, Gamemaster
Internal ops, scoring, gamification
Kanban Task Board
Five-column workflow for task management:
Inbox - Captured thoughts and incoming tasks
Assigned - Tasks assigned to specific agents
Working - Tasks actively being worked on
Review - Tasks awaiting review/approval
Done - Completed tasks
Features
Drag-and-drop task movement between columns
Quick capture for rapid thought capture
Task detail sheets with full metadata
Real-time updates via WebSocket
"Squirrel" tags for captured thoughts
Scoring System
Agents earn or lose points based on feedback sentiment in journal entries.
Credits (+5 points each)
Positive feedback patterns:
"perfect", "thanks", "good job"
"works great", "excellent", "nice"
"ship it", "commit", "approved", "done"
Debits (-10 points each)
Negative feedback patterns:
"that's not right", "you missed", "why didn't you"
"broken", "doesn't work", "failed", "not working"
"you forgot", "should have", "lazy"
Status Tiers
Balance
Status
Meaning
< 0
In Debt
Needs improvement
0 - 49
Standard
Meeting expectations
50 - 99
Good
Exceeding expectations
100+
Autonomous
High trust, minimal oversight
Note: The asymmetric scoring (-10 for failures vs +5 for successes) reflects that avoiding mistakes is weighted more heavily than earning praise.
Architecture
Data Flow
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ Local API │────▶│ Dashboard │◀────│ Cloudflare KV │
│ (port 8891) │ │ (index.html) │ │ (fallback) │
└─────────────────┘ └─────────────────┘ └─────────────────┘
│ │
│ ▼
│ ┌─────────────────┐
└──────────────▶│ WebSocket │
│ (live updates) │
└─────────────────┘
API Endpoints
Endpoint
Purpose
/status Board state, agents, tasks
/alerts Critical alerts
/ledger Venture ledger data
/transactions P&L transaction history
/scores Agent score data
/inbox Task bus for injected tasks
/journal Journal entries for sentiment
Fallback Chain
Local API (localhost:8891) - Primary, live workspace data
Cloudflare Worker - Remote KV store backup
Static JSON files - Bundled fallback data
Usage
Quick Capture
Click the + Capture button in the header
Type your thought or task
Press Enter to add
Task appears in Inbox as a "squirrel" (captured thought)
Managing Tasks
Drag cards between columns to change status
Click cards to open the detail sheet
Tasks in Done column auto-limit to 5 most recent
P&L Scoreboard
Click agent rows (Claude Code, OpenClaw) to expand/collapse
Expansion state persists across page refreshes
Categories show subtotals for quick analysis
Keyboard Shortcuts
Escape - Close detail sheet or capture bar
Design Philosophy
Transparency - All agent activity is visible and tracked
Accountability - P&L scoring creates clear performance metrics
Gamification - Streaks, scores, and status tiers motivate improvement
Real-time - WebSocket updates keep dashboard current
Resilience - Multiple fallback data sources ensure availability
Mission Control - Command your agent swarm with clarity and confidence.