Agents

AI development agents that execute planned work, create branches, and open pull requests.

Enhance integrates with AI development agents to execute planned work autonomously or semi-autonomously. Agents create branches, write code, and open pull requests.

Supported providers

Provider	Status
Cursor	Primary integration
Claude	Supported
Copilot	Supported
Devin	Supported
Bolt	Supported
Replit	Supported
Windsurf	Supported
Lovable	Supported
v0	Supported

Agent runs

An agent run is a planned batch of work. It connects to the execution plan and tracks progress across multiple issues.

Run lifecycle

Draft — Run created, issues selected
Planning — AI generates execution plan with overlap analysis
Ready — Plan reviewed and approved
Executing — Agents actively working on issues
Completed — All issues processed

Run properties

Each run tracks:

Name and description — What the run is for
Target branch — Base branch for agent work
Auto-create PR — Whether to open PRs automatically
Planning data — Overlap analysis, AI batches, reasoning, flow chart
Stats — Total, completed, and failed issues
Timeline — Start time, completion time, duration

Agent executions

Each execution is an individual agent instance working on one or more issues:

Provider and external ID — Which agent service and its identifier
Status — Creating, Running, Completed, Failed, Stopped
Prompt — Instructions sent to the agent
Source repo and branch — Where to read code from
Target branch and PR config — Where to push changes
Results — PR URL, number, and summary of changes
Conversation log — Full transcript of agent interaction
Token usage — Cost tracking

Execution flow

Issues in a run are executed respecting the planned order:

Issues with no dependencies start first
As issues complete, dependent issues become unblocked
Issues with file overlaps are tracked for potential conflicts
Each issue tracks its status: Pending, Executing, Completed, Failed

Dashboard

The agent dashboard provides:

Active runs — Real-time status with live timers
Run history — Past runs with outcomes
Run detail — Per-issue status, expandable with agent console output
Execution flow canvas — Visual execution order
Cursor status indicator — Real-time connection to Cursor agents