Agent Commands
Spawn, monitor, and control autonomous AI coding agents.
pan work issue
Spawn an agent for a Linear or GitHub issue.
# Spawn an agent for a Linear issue
pan work issue MIN-123
# Spawn an agent for a GitHub issue
pan work issue 42
- Creates a workspace (git worktree) if it doesn’t exist
- Spawns an autonomous agent in a tmux session
- The agent reads the issue, explores the codebase, and implements the solution
pan work status
List all running agents and their status.
Shows:
- Agent session ID
- Issue number and title
- Current status
- Runtime duration
- Workspace location
The dashboard’s Agents page is the visual counterpart to this command. A metric strip across the top tallies how many agents are running, how many are stuck, and the rolling cost and token spend over the last 24 hours, with each agent rendered as a card showing its model, runtime, and last activity.
Use pan work status from the terminal for a quick check, and the Agents page when you want to watch many agents at once and spot stuck ones at a glance.
pan work tell
Send a message to a running agent.
# Send a message to an agent (handles Enter key automatically)
pan work tell min-123 "Please also add tests"
- Sends your message to the agent’s tmux session
- The agent receives it and responds accordingly
- Useful for providing feedback, clarification, or guidance
pan work kill
Stop a running agent.
Terminates the agent’s tmux session. The workspace and any uncommitted work remain intact.
pan work pending
Show work completed and awaiting review.
Lists issues where agents have finished work but haven’t been approved yet.
pan work approve
Approve agent work, merge MR, and update issue tracker.
This command:
- Spawns a merge-agent specialist
- Reviews the changes
- Merges the pull/merge request
- Updates the issue status
pan work list
List issues from configured trackers.
Shows issues from Linear, GitHub, or GitLab based on your configuration.
pan work triage
Triage issues from secondary tracker.
Process issues from a secondary tracker (e.g., Rally) and create equivalent issues in your primary tracker.
pan work reopen
Reopen a closed issue and re-run planning.
Reopens the issue in your tracker and restarts the planning session.
pan work request-review
Request re-review after fixing feedback.
pan work request-review min-123 -m "Fixed: added tests and removed duplication"
pan work wipe
Deep wipe: completely reset all state for an issue.
# Clean up agents, state, Linear status
pan work wipe min-123
# Also delete workspace
pan work wipe min-123 -w
# Skip confirmation
pan work wipe min-123 -y -w
- Kills any running agents
- Cleans up state files
- Updates issue tracker status
- Optionally removes the workspace
pan work recover
Recover crashed agents.
# Recover a specific agent
pan work recover min-123
# Recover all crashed agents
pan work recover --all
pan specialists
Manage specialist agents (review, test, inspect, UAT, merge).
# Check status of all specialists
pan specialists status
# Wake a specialist for an issue
pan specialists wake review-agent --issue MIN-123
# View specialist queue
pan specialists queue review-agent
# Clear a specialist's queue
pan specialists clear-queue review-agent
# Reset a specialist's state
pan specialists reset review-agent
# View specialist logs
pan specialists logs review-agent
pan remote
Manage remote workspaces via Fly.io.
# Initialize Fly.io provider
pan remote init
# Configure remote environment
pan remote setup
# Check remote workspace status
pan remote status
# View resource allocation
pan remote resources
Next Steps
