AI Agents
Deploy AI agents that analyze code, generate solutions, monitor issues, and automate development tasks within your project.
Agent Types
| Agent | Purpose |
|---|---|
| TicketAgent | Creates, triages, and manages tickets from conversations |
| TaskAgent | Breaks down objectives into subtasks with effort estimates |
| CodeAgent | Generates, reviews, and refactors code autonomously |
| PlannerAgent | Creates strategic plans with milestones and risk assessments |
Getting Started
- Navigate to your project dashboard
- Click Agents in the sidebar
- Choose an agent template (Developer, Reviewer, Tester, etc.)
- Configure autonomy level, permissions, and integration points
- Deploy and monitor activity in the dashboard
Autonomy Levels
| Level | Behavior |
|---|---|
| Conservative | Always asks for approval before acting |
| Balanced | Acts autonomously on low-risk decisions |
| Aggressive | Acts on most decisions, asks only for critical ones |
| Autonomous | Fully autonomous — no approval gates |
Safety Controls
Agents operate under strict safety controls. Tool calls classified as dangerous require explicit user approval. Execution loops are capped at 50 iterations, and rate limits are enforced by subscription tier.
Related
- Agent SDK Reference — Build agents programmatically
- Runtime Engine — Task execution internals
- AI Monitoring — Track agent LLM usage