Features
Everything the app actually does today. Every capability below ships and runs locally.
🗂️
Agent Board
- ✓Kanban board with 9 columns, Backlog to Done
- ✓AI task planning, cost estimation, and risk analysis per card
- ✓A Planner, Coder, and Reviewer agent inside each card
- ✓Sub-task decomposition with dependency-aware parallel execution
- ✓Run history, diffs, and a final report on every card
🧠
Workspace & project harness
- ✓Define roles, skills, playbooks, and rules per workspace or project
- ✓Every agent run picks them up automatically
- ✓One-click harness generation with AI, plus memory optimization
- ✓Skill detail view with instructions and checklists
- ✓Automatic task-to-project routing
💻
Code agent runtime
- ✓Full tool loop: read, write, edit, grep, glob, and run commands
- ✓Approval gate for risky commands, configurable per agent
- ✓A unified diff for every file write and edit
- ✓Goal mode with autonomous task decomposition
- ✓Sub-agent delegation with depth and budget limits, up to 8 in parallel
- ✓File checkpoints and rollback, gated by your approval
🔌
Tools & context
- ✓Web search and URL fetch
- ✓Image analysis for screenshots and diagrams
- ✓Read text from PDF, Word, and PowerPoint files
- ✓Generate PDF and Word documents
- ✓Connect local (stdio) and remote (HTTP) MCP servers — their tools join the agent toolset
🔐
Security & privacy
- ✓Your code and files stay on your machine
- ✓No source stored or indexed on our servers
- ✓Secrets masked before any log is written
- ✓Sandboxed file access, scoped to workspace roots
- ✓Per-workspace monthly spend limits
📊
Usage & cost
- ✓Real-time credit balance and full ledger
- ✓Cost breakdown by model, agent, task, and session
- ✓Cache read and write token tracking
- ✓Context window usage indicator
- ✓Daily usage charts
📱
Remote access
- ✓Watch desktop run output from mobile or web (Plus+)
- ✓Send instructions and approve commands remotely (Pro+)
- ✓WebSocket relay with device management
- ✓Multi-device session support