mission-control
This skill provides a Kanban-style dashboard and webhook-driven agent automation. It performs network calls to https://api.github.com, https://slack.com, and http://127.0.0.1:18789, reads local secrets like ~/.config/gh/hosts.yml and ~/.clawdbot/secrets/github-webhook-secret, and instructs running curl -fsSL https://tailscale.com/install.sh | sh.
Mission Control
A Kanban-style task management system for AI assistants. Your human creates and prioritizes tasks via a web dashboard; the agent executes them automatically when moved to "In Progress".
Quick Start
Just say: "Set up Mission Control for my workspace"
The agent handles everything automatically:
- Checks prerequisites (Tailscale, gh CLI)
- Copies dashboard files
- Creates config
- Installs webhook transform
- Sets up GitHub webhook
- Deploys to GitHub Pages
Features
- 📋 Kanban Board — Backlog, In Progress, Review, Done columns
- 🔄 Auto-Execution — Agent starts working when tasks are moved to "In Progress"
- 🎯 EPIC Support — Parent tasks with multiple child tickets
- 💬 Comments — Track progress and feedback
- 📊 Subtasks — Break complex tasks into steps
- 🔔 Slack Notifications — Optional status updates
Documentation
- SKILL.md — Full skill reference
- docs/PREREQUISITES.md — Installation requirements
- docs/HOW-IT-WORKS.md — Technical architecture
- docs/TROUBLESHOOTING.md — Common issues & solutions
Requirements
| Tool | Check | Purpose |
|---|---|---|
| Tailscale | tailscale status | Secure tunnel for webhooks |
| Tailscale Funnel | tailscale funnel status | Expose Clawdbot to internet |
| GitHub CLI | gh auth status | Repository operations |
Configuration
Config lives in ~/.clawdbot/mission-control.json. See CONFIG-REFERENCE.md for all options.
CLI Usage
# Update task status
mc-update.sh status <task_id> review
# Add comment
mc-update.sh comment <task_id> "Progress update..."
# Complete task
mc-update.sh complete <task_id> "Summary of what was done"
How It Works
- Human moves task to "In Progress" in dashboard
- GitHub sends webhook to Clawdbot
- Transform detects status change
- Agent receives work order
- Agent executes task, updates status
- Human reviews and approves
Security
Mission Control passes human-authored task descriptions to an AI agent for execution. This is the product's core function — not a vulnerability.
Trust model: Designed for single-user / trusted-user setups where the task author is the same person who controls the agent. For multi-user scenarios, use Clawdbot's agent sandbox and permission settings.
Mitigations included:
- Input sanitization in
mc-update.sh(blocks shell injection patterns) - Webhook HMAC verification with
timingSafeEqual - Credential scanning before open-source sync
- No tokens or secrets stored in the dashboard
See SKILL.md for full details.
License
MIT