Synthèse automatique
Fournit une référence complète des commandes CLI de Gastown (gt) utilisées pour gérer les services, l'orchestration, les workers, l'espace de travail, la communication et les diagnostics.
Points forts
- Couverture exhaustive de tous les groupes de commandes
- inclut les drapeaux et exemples
- conçu pour une utilisation interne par Claude Code.
Limites
- Ne couvre que l'outil Gastown
- aucune intégration externe
- nécessite la connaissance des concepts Gastown.
Utilisez cette référence lorsque vous devez exécuter des commandes CLI Gastown pour la gestion des services ou l'orchestration.
Ne l'utilisez pas pour des commandes shell générales ou d'autres outils.
Analyse de sécurité
SûrThis skill is purely a reference document for 'gt' CLI commands; it contains no executable payloads, destructive commands, or exfiltration risks.
Aucun point d'attention détecté
Comment cette analyse est faiteExemples
Run gt up with the --restore flag to bring up all Gastown services and also restore crew and polecats.Execute gt stop --all --graceful to kill all polecat sessions across all rigs, attempting graceful shutdown first.Create a new convoy with the associated molecule ID 'molecule-123' using gt convoy create --molecule molecule-123.Gastown CLI Commands Reference
Internal Reference for Claude Code - You run all these commands for the user.
Complete reference for gt (gastown) CLI commands. All commands support --help for detailed usage.
Command Groups
Commands are organized into functional groups:
| Group | Description | |-------|-------------| | services | Service lifecycle (up, down, stop) | | orchestration | Work coordination (convoy, sling, broadcast) | | workers | Worker management (polecat, crew, dog) | | workspace | Workspace operations (init, rig, worktree) | | communication | Inter-agent messaging (mail, escalate, handoff) | | diagnostics | Status and debugging (status, prime, whoami, version) |
Service Lifecycle Commands
gt up
Bring up all Gas Town services (idempotent boot command).
Syntax:
gt up [flags]
Flags:
| Flag | Description |
|------|-------------|
| -q, --quiet | Only show errors |
| --restore | Also restore crew (from settings) and polecats (from hooks) |
Services Started:
- Daemon - Go background process that pokes agents
- Deacon - Health orchestrator (monitors Mayor/Witnesses)
- Mayor - Global work coordinator
- Witnesses - Per-rig polecat managers
- Refineries - Per-rig merge queue processors
Examples:
gt up # Start all infrastructure
gt up --quiet # Start silently, show only errors
gt up --restore # Also start crew and polecats with pinned work
Common Use Cases:
- Starting your Gas Town workspace after boot
- Recovering after a system restart
- Ensuring all services are running before starting work
gt down
Stop all Gas Town infrastructure services.
Syntax:
gt down [flags]
Flags:
| Flag | Description |
|------|-------------|
| -q, --quiet | Only show errors |
| -f, --force | Force kill without graceful shutdown |
| -a, --all | Also kill the tmux server |
What it stops:
- Witnesses (per-rig polecat managers)
- Mayor (global work coordinator)
- Boot (Deacon's watchdog)
- Deacon (health orchestrator)
- Daemon (Go background process)
Note: Polecats are NOT stopped by gt down. Use gt stop --all for emergency polecat shutdown.
Examples:
gt down # Graceful shutdown of infrastructure
gt down --force # Force immediate shutdown
gt down --all # Also kill tmux server
Common Use Cases:
- Shutting down for the night
- Preparing for maintenance
- Clean restart of services
gt stop
Emergency stop for polecat sessions.
Syntax:
gt stop [flags]
Flags:
| Flag | Description |
|------|-------------|
| --all | Stop all sessions across all rigs (required if no --rig) |
| --rig <name> | Stop all sessions in a specific rig |
| --graceful | Try graceful shutdown before force kill |
Note: Must specify either --all or --rig. This is for emergency polecat shutdown only.
Examples:
gt stop --all # Kill ALL polecat sessions across all rigs
gt stop --rig wyvern # Kill all sessions in the wyvern rig
gt stop --all --graceful # Try graceful shutdown first
Orchestration Commands
gt convoy
Track batches of work across rigs. A convoy is a persistent tracking unit that monitors related issues.
Syntax:
gt convoy [subcommand] [flags]
Parent Command Flags:
| Flag | Description |
|------|-------------|
| -i, --interactive | Launch interactive TUI dashboard |
Subcommands:
| Subcommand | Description |
|------------|-------------|
| create | Create a new convoy |
| list | List convoys |
| status | Show convoy status |
| add | Add issues to an existing convoy |
Flags for convoy create:
| Flag | Description |
|------|-------------|
| --molecule string | Associated molecule ID |
| --notify | Address to notify on completion (default: mayor/) |
Flags for convoy list:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
| --status string | Filter by status (open, closed) |
| --all | Show all convoys (open and closed) |
Flags for convoy status:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
Examples:
gt convoy create "Deploy v2.0" gt-abc bd-xyz # Create convoy tracking issues
gt convoy create "Release prep" gt-abc --notify # Notify mayor on completion
gt convoy list # List open convoys
gt convoy list --all # List all convoys
gt convoy status hq-cv-abc # Detailed convoy status
gt convoy status 1 # Status by number (from list)
gt convoy add hq-cv-abc gt-new-issue # Add issue to convoy
gt convoy -i # Interactive TUI
Common Use Cases:
- Tracking batches of related work across rigs
- Monitoring progress of parallel tasks
- Getting notified when all tracked issues complete
gt sling
Sling work onto an agent's hook and start working immediately. THE unified work dispatch command.
Syntax:
gt sling <bead-or-formula> [target] [flags]
Target Resolution:
gt sling gt-abc- Self (current agent)gt sling gt-abc crew- Crew worker in current riggt sling gt-abc greenplace- Auto-spawn polecat in riggt sling gt-abc greenplace/Toast- Specific polecatgt sling gt-abc mayor- Mayorgt sling gt-abc deacon/dogs- Auto-dispatch to idle doggt sling gt-abc deacon/dogs/alpha- Specific dog
Flags:
| Flag | Description |
|------|-------------|
| -s, --subject string | Context subject for the work |
| -m, --message string | Context message for the work |
| -n, --dry-run | Show what would be done |
| --on string | Apply formula to existing bead (implies wisp scaffolding) |
| --var stringArray | Formula variable (key=value), can be repeated |
| -a, --args string | Natural language instructions for the executor |
| --naked | No-tmux mode: assign work but skip session creation |
| --create | Create polecat if it doesn't exist |
| --molecule string | Molecule workflow to instantiate on the bead |
| --force | Force spawn even if polecat has unread mail |
| --account string | Claude Code account handle to use |
| -q, --quality string | Polecat workflow quality level (basic|shiny|chrome) |
| --no-convoy | Skip auto-convoy creation for single-issue sling |
Examples:
gt sling gt-abc greenplace # Auto-spawn polecat in rig
gt sling gt-abc greenplace --quality=shiny # Use shiny workflow
gt sling gt-abc --args "patch release" # With natural language args
gt sling mol-review --on gt-abc mayor # Apply formula to existing work
gt sling gt-abc greenplace --naked # No-tmux mode (manual start)
gt sling gt-abc deacon/dogs --create # Dispatch to dog pool
Compare Related Commands:
gt hook <bead>- Just attach (no action)gt sling <bead>- Attach + start now (keep context)gt handoff <bead>- Attach + restart (fresh context)
gt broadcast
Send a message to all workers.
Syntax:
gt broadcast <message> [flags]
Flags:
| Flag | Description |
|------|-------------|
| --rig string | Only broadcast to workers in this rig |
| --all | Include all agents (mayor, witness, etc.), not just workers |
| --dry-run | Show what would be sent without sending |
Examples:
gt broadcast "Check your mail"
gt broadcast --rig gastown "New priority work available"
gt broadcast --all "System maintenance in 5 minutes"
gt broadcast --dry-run "Test message"
Common Use Cases:
- Notifying workers of priority changes
- Announcing system maintenance
- Coordinating mass check-ins
gt escalate
Escalate an issue to the human overseer.
Syntax:
gt escalate <topic> [flags]
Flags:
| Flag | Description |
|------|-------------|
| -s, --severity string | Severity level: CRITICAL, HIGH, or MEDIUM (default: MEDIUM) |
| -m, --message string | Additional details about the escalation |
| -n, --dry-run | Show what would be done without executing |
Severity Levels: | Level | Priority | Description | |-------|----------|-------------| | CRITICAL | P0 | System-threatening, immediate attention required | | HIGH | P1 | Important blocker, needs human soon | | MEDIUM | P2 | Standard escalation, human attention at convenience |
Examples:
gt escalate "Database migration failed"
gt escalate -s CRITICAL "Data corruption detected in user table"
gt escalate -s HIGH "Merge conflict cannot be resolved automatically"
gt escalate -s MEDIUM "Need clarification on API design" -m "Details here..."
Common Use Cases:
- Reporting blockers that require human decision
- Escalating security or data issues
- Requesting clarification on ambiguous requirements
gt swarm
DEPRECATED: Use gt convoy instead.
Create swarm work units (legacy command).
Syntax:
gt swarm [flags]
Note: This command is deprecated. All swarm functionality has been migrated to the gt convoy command. Please use gt convoy for all new work.
gt mol
Agent-specific molecule workflow operations. Operates on YOUR hook and YOUR attached molecules.
Syntax:
gt mol [subcommand] [flags]
Aliases: molecule
Subcommands:
| Subcommand | Description |
|------------|-------------|
| status | Show what's on your hook (alias: gt hook) |
| current | Show what you should be working on |
| progress | Show execution progress through steps |
| step done | Complete current step (auto-continues) |
| attach | Attach molecule to your hook |
| detach | Detach molecule from your hook |
| burn | Discard attached molecule (no record) |
| squash | Compress to digest (permanent record) |
Examples:
gt hook # Show what's on my hook
gt mol current # What should I be working on?
gt mol progress # Show step progress
gt mol step done # Complete current step
gt mol attach mol-abc # Attach molecule to hook
gt mol detach # Detach molecule
gt mol burn # Discard without record
gt mol squash # Compress to digest
Related Commands:
gt sling mol-xxx target- Pour formula + sling to agentgt formula list- List available formulas
gt formula
Manage workflow formulas - reusable molecule templates.
Syntax:
gt formula [subcommand] [flags]
Aliases: formulas
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List available formulas from all search paths |
| show <name> | Display formula details (steps, variables) |
| run <name> | Execute a formula (pour and dispatch) |
| create <name> | Create a new formula template |
Search Paths (in order):
.beads/formulas/(project)~/.beads/formulas/(user)$GT_ROOT/.beads/formulas/(orchestrator)
Examples:
gt formula list # List all formulas
gt formula show shiny # Show formula details
gt formula run shiny --pr=123 # Run formula on PR #123
gt formula create my-workflow # Create new formula template
Worker Management Commands
gt polecat
Manage polecat (transient worker) agents. Aliases: cat, polecats.
Syntax:
gt polecat <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list [rig] | List polecats in a rig or all rigs |
| add <rig> <name> | Create a new polecat |
| remove <rig/polecat> | Remove polecat(s) |
| status <rig/polecat> | Show detailed polecat status |
| sync <rig/polecat> | Sync beads for polecat worktree |
| done <rig/polecat> | Mark polecat as done with work |
| reset <rig/polecat> | Force reset polecat to idle state |
| gc <rig> | Garbage collect stale polecat branches |
| nuke <rig/polecat> | Completely destroy polecat (session, worktree, branch, bead) |
| git-state <rig/polecat> | Show git state for pre-kill verification |
| check-recovery <rig/polecat> | Check if polecat needs recovery vs safe to nuke |
Flags for polecat list:
| Flag | Description |
|------|-------------|
| --all | List polecats in all rigs |
| --json | Output as JSON |
Flags for polecat remove:
| Flag | Description |
|------|-------------|
| -f, --force | Force removal, bypassing checks |
| --all | Remove all polecats in the rig |
Flags for polecat sync:
| Flag | Description |
|------|-------------|
| --all | Sync all polecats in the rig |
| --from-main | Pull only, no push |
Flags for polecat gc:
| Flag | Description |
|------|-------------|
| --dry-run | Show what would be deleted without deleting |
Flags for polecat nuke:
| Flag | Description |
|------|-------------|
| --all | Nuke all polecats in the rig |
| --dry-run | Show what would be nuked without doing it |
| -f, --force | Force nuke, bypassing all safety checks (LOSES WORK) |
Flags for polecat status, git-state, check-recovery:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
Examples:
gt polecat list greenplace # List polecats in rig
gt polecat list --all # List across all rigs
gt polecat add greenplace Toast # Create polecat Toast in greenplace
gt polecat remove greenplace/Toast # Remove single polecat
gt polecat remove greenplace --all # Remove all polecats in rig
gt polecat status greenplace/Toast # Show detailed status
gt polecat sync greenplace/Toast # Sync beads for polecat
gt polecat gc greenplace --dry-run # Preview branch cleanup
gt polecat nuke greenplace/Toast # Nuclear cleanup (post-merge)
gt polecat nuke greenplace --all # Nuke all polecats
Common Use Cases:
- Spawning workers for parallel tasks
- Cleaning up after convoy completion
- Inspecting worker status
- Post-merge cleanup with nuke
gt crew
Manage crew (user-managed persistent workspaces within a rig).
Unlike polecats which are witness-managed and transient, crew workers are:
- Persistent: Not auto-garbage-collected
- User-managed: Overseer controls lifecycle
- Long-lived identities: Recognizable names like dave, emma, fred
- Gas Town integrated: Mail, handoff mechanics work
- Tmux optional: Can work in terminal directly
Syntax:
gt crew <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List crew workspaces with status |
| add <name...> | Create new crew workspace(s) |
| at [name] | Attach to crew workspace session (alias: attach) |
| remove <name...> | Remove crew workspace(s) |
| refresh <name> | Context cycling with mail-to-self handoff |
| restart [name...] | Kill and restart session fresh (alias: rs) |
| status [name] | Show detailed workspace status |
| rename <old> <new> | Rename a crew workspace |
| pristine [name] | Sync crew workspaces with remote |
| start [name...] | Start crew workspace(s), creates if needed |
| next | Switch to next crew session in same rig |
| prev | Switch to previous crew session in same rig |
Flags for crew list:
| Flag | Description |
|------|-------------|
| --rig string | Filter by rig name |
| --json | Output as JSON |
Flags for crew add:
| Flag | Description |
|------|-------------|
| --rig string | Rig to create crew workspace in |
| --branch | Create a feature branch (crew/<name>) |
Flags for crew at:
| Flag | Description |
|------|-------------|
| --rig string | Rig to use |
| --no-tmux | Just print directory path |
| -d, --detached | Start session without attaching |
| --account string | Claude Code account handle to use |
Flags for crew remove:
| Flag | Description |
|------|-------------|
| --rig string | Rig to use |
| --force | Force remove (skip safety checks) |
Flags for crew restart:
| Flag | Description |
|------|-------------|
| --rig string | Rig to use (filter when using --all) |
| --all | Restart all running crew sessions |
| --dry-run | Show what would be restarted without restarting |
Flags for crew start:
| Flag | Description |
|------|-------------|
| --rig string | Rig to use |
| --account string | Claude Code account handle to use |
Examples:
gt crew list # List crew workers
gt crew add dave # Create crew worker 'dave'
gt crew add murgen croaker goblin # Create multiple at once
gt crew at dave # Attach to dave's session
gt crew at dave --detached # Start session without attaching
gt crew remove dave # Remove crew worker
gt crew status dave # Show detailed status
gt crew refresh dave -m "Working on gt-123" # Cycle with handoff
gt crew restart dave # Kill and restart fresh
gt crew restart --all # Restart all crew sessions
gt crew pristine dave # Sync workspace with remote
gt crew start joe # Start joe (creates if needed)
gt crew next # Cycle to next worker
gt crew prev # Cycle to previous
Common Use Cases:
- Setting up long-running development agents
- Managing personal AI assistants per rig
- Switching between crew workers
gt dog
Manage dogs (Deacon's helper workers for infrastructure tasks).
Dogs are reusable helper workers managed by the Deacon for infrastructure and cleanup tasks. Unlike polecats (single-rig, ephemeral), dogs handle cross-rig infrastructure work with worktrees into each rig.
Syntax:
gt dog <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List all dogs in the kennel |
| add <name> | Create a new dog in the kennel |
| remove <name>... | Remove dogs from the kennel |
| call [name] | Wake idle dog(s) for work |
| status [name] | Show detailed dog status |
Flags for dog list:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
Flags for dog remove:
| Flag | Description |
|------|-------------|
| -f, --force | Force removal even if working |
| --all | Remove all dogs |
Flags for dog call:
| Flag | Description |
|------|-------------|
| --all | Wake all idle dogs |
Flags for dog status:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
Examples:
gt dog list # List all dogs
gt dog add alpha # Create dog 'alpha'
gt dog add bravo # Create dog 'bravo'
gt dog call alpha # Wake specific dog
gt dog call --all # Wake all idle dogs
gt dog call # Wake one idle dog
gt dog status alpha # Show dog status
gt dog status # Show pack summary
gt dog remove alpha # Remove dog
gt dog remove alpha bravo # Remove multiple
gt dog remove --all # Remove all dogs
gt dog remove alpha --force # Force remove working dog
Common Use Cases:
- Cross-rig infrastructure tasks
- Cleanup and maintenance operations
- Deacon-managed background work
Merge Queue Commands
gt mq
Manage the merge queue (refined work integration).
Syntax:
gt mq [subcommand] [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List items in merge queue |
| submit <branch> | Submit a branch to the queue |
| retry <item> | Retry a failed queue item |
| reject <item> | Reject an item from the queue |
| status | Show queue status |
Flags for mq list:
| Flag | Description |
|------|-------------|
| --rig string | Target rig |
| --json | Output as JSON |
| --all | Show all statuses |
Examples:
gt mq list # List queue items
gt mq submit feature/my-branch # Submit to queue
gt mq retry mq-001 # Retry failed item
gt mq reject mq-002 # Reject an item
gt mq status # Show queue status
gt refinery
Manage refinery (merge processor) agents.
Syntax:
gt refinery <subcommand> [rig] [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| status [rig] | Show refinery status |
| start <rig> | Start the refinery for a rig |
| stop <rig> | Stop the refinery for a rig |
| attach <rig> | Attach to refinery session |
| restart <rig> | Restart the refinery |
| queue [rig] | Show the merge queue |
| claim <branch> | Claim a branch for processing |
| release <branch> | Release a claimed branch |
Flags:
| Flag | Description |
|------|-------------|
| --rig string | Target rig |
| --json | Output as JSON |
Examples:
gt refinery status # Show all refinery status
gt refinery status myrig # Show refinery for specific rig
gt refinery start myrig # Start refinery for rig
gt refinery stop myrig # Stop refinery for rig
gt refinery attach myrig # Attach to refinery session
gt refinery restart myrig # Restart refinery
gt refinery queue # Show merge queue
gt refinery claim polecat/toast # Claim branch for processing
gt refinery release polecat/toast # Release claimed branch
Common Use Cases:
- Monitoring merge progress
- Checking for stalled merges
- Managing refinery lifecycle
- Manual claim/release for conflict resolution
Workspace Commands
gt init
Initialize a new Gas Town workspace.
Syntax:
gt init [flags]
Flags:
| Flag | Description |
|------|-------------|
| --name string | Town name (default: directory name) |
| --path string | Path to create workspace (default: ~/gt) |
| --force | Overwrite existing workspace |
Examples:
gt init # Initialize in ~/gt
gt init --name mytown # Custom name
gt init --path /custom/path # Custom location
gt rig
Manage rigs (isolated project workspaces).
Syntax:
gt rig <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List all rigs |
| add <name> | Add a new rig |
| remove <name> | Remove a rig |
| show <name> | Show rig details |
Flags for rig add:
| Flag | Description |
|------|-------------|
| --repo string | Git repository URL |
| --branch string | Default branch (default: main) |
| --prefix string | Beads prefix for the rig |
Examples:
gt rig list # List all rigs
gt rig add myproject --repo git@github.com:user/repo.git
gt rig show gastown # Show rig details
gt rig remove oldproject # Remove a rig
gt worktree
Create worktrees in other rigs for cross-rig work.
Syntax:
gt worktree <rig> [flags]
gt worktree list
gt worktree remove <rig>
Subcommands:
| Subcommand | Description |
|------------|-------------|
| list | List all cross-rig worktrees owned by current crew member |
| remove <rig> | Remove a cross-rig worktree |
Flags:
| Flag | Description |
|------|-------------|
| --no-cd | Just print path (don't print cd command) |
| -f, --force | Force remove even with uncommitted changes (for remove) |
Examples:
gt worktree beads # Create worktree in beads rig
gt worktree gastown # Create worktree in gastown rig
gt worktree beads --no-cd # Just print the path
gt worktree list # List your worktrees
gt worktree remove beads # Remove beads worktree
gt worktree remove beads --force # Force remove
Common Use Cases:
- Working on multiple rigs from a single identity
- Cross-rig dependency work
- Collaborative fixes across projects
Communication Commands
gt mail
Internal mail system for agent communication. Messages are stored as beads.
Mail Routing:
┌─────────────────────────────────────────────────────┐
│ Town (.beads/) │
│ ┌─────────────────────────────────────────────┐ │
│ │ Mayor Inbox │ │
│ │ └── mayor/ │ │
│ └─────────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────────┐ │
│ │ <rig>/ (rig mailboxes) │ │
│ │ ├── witness ← <rig>/witness │ │
│ │ ├── refinery ← <rig>/refinery │ │
│ │ ├── <polecat> ← <rig>/<polecat> │ │
│ │ └── crew/<name> ← <rig>/crew/<name> │ │
│ └─────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────┘
Syntax:
gt mail <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| inbox | Show inbox for current identity |
| send | Send a message |
| read <id> | Read a specific message |
| check | Check for new mail |
| reply <id> | Reply to a message |
| archive <id> | Archive a message |
| peek | Show preview of first unread message |
| thread <id> | View a message thread |
| search <query> | Search messages by content |
| clear | Clear all messages from inbox |
Address Formats:
| Address | Recipient |
|---------|-----------|
| mayor/ | Mayor inbox |
| <rig>/witness | Rig's Witness |
| <rig>/refinery | Rig's Refinery |
| <rig>/<polecat> | Polecat (e.g., greenplace/Toast) |
| <rig>/crew/<name> | Crew worker (e.g., greenplace/crew/max) |
| --human | Human overseer (special) |
Flags for mail inbox:
| Flag | Description |
|------|-------------|
| --identity string | Override identity |
| --unread | Show only unread messages |
| --json | Output as JSON |
Flags for mail send:
| Flag | Description |
|------|-------------|
| --to string | Recipient address |
| --subject string | Subject line |
| --body string | Message body |
| --priority string | Priority (low, normal, high, urgent) |
Examples:
gt mail inbox # Check inbox
gt mail inbox --identity mayor # Check Mayor's inbox
gt mail send --to mayor/ --subject "Update" --body "Work complete"
gt mail send --to greenplace/witness --subject "Patrol" --body "Process work"
gt mail read msg-abc123 # Read message
gt mail reply msg-abc123 # Reply to message
gt mail check --inject # Check and inject mail into context
gt mail thread msg-abc123 # View conversation thread
gt mail search "merge conflict" # Search for messages
gt nudge
Send a message to an agent session reliably. This is the canonical way to communicate with running Claude sessions.
Syntax:
gt nudge <target> [message] [flags]
Target Formats:
| Target | Session |
|--------|---------|
| mayor | gt-mayor |
| deacon | gt-deacon |
| witness | gt-<rig>-witness (current rig) |
| refinery | gt-<rig>-refinery (current rig) |
| <rig>/<polecat> | gt-<rig>-polecat-<name> |
| channel:<name> | All members of a named channel |
Flags:
| Flag | Description |
|------|-------------|
| -m, --message string | Message to send |
| -f, --force | Send even if target has DND enabled |
Examples:
gt nudge greenplace/furiosa "Check your mail and start working"
gt nudge mayor "Status update requested"
gt nudge witness "Check polecat health"
gt nudge deacon session-started
gt nudge channel:workers "New priority work available"
Note: If target has DND (Do Not Disturb) enabled, nudge is skipped unless --force is used.
gt peek
View recent output from an agent session. The companion to gt nudge for agent communication.
Syntax:
gt peek <rig/polecat> [count] [flags]
Flags:
| Flag | Description |
|------|-------------|
| -n, --lines int | Number of lines to capture (default 100) |
Examples:
gt peek greenplace/furiosa # Last 100 lines (default)
gt peek greenplace/furiosa 50 # Last 50 lines
gt peek greenplace/furiosa -n 200 # Last 200 lines
Common Use Cases:
- Checking what an agent is currently doing
- Debugging stuck agents
- Reviewing recent agent output
The nudge/peek pair:
gt nudge- send messages TO a session (reliable delivery)gt peek- read output FROM a session (capture-pane wrapper)
gt handoff
Hand off to a fresh session. Work continues from hook. The canonical way to end any agent session.
Handles all roles:
- Mayor, Crew, Witness, Refinery, Deacon: Respawns with fresh Claude instance
- Polecats: Calls
gt done --exit DEFERRED(Witness handles lifecycle)
Syntax:
gt handoff [bead-or-role] [flags]
Flags:
| Flag | Description |
|------|-------------|
| -w, --watch | Switch to new session (for remote handoff, default: true) |
| -n, --dry-run | Show what would be done without executing |
| -s, --subject string | Subject for handoff mail (optional) |
| -m, --message string | Message body for handoff mail (optional) |
| -c, --collect | Auto-collect state (status, inbox, beads) into handoff message |
Examples:
gt handoff # Hand off current session
gt handoff gt-abc # Hook bead, then restart
gt handoff gt-abc -s "Fix it" # Hook with context, then restart
gt handoff -s "Context" -m "Notes" # Hand off with custom message
gt handoff -c # Collect state into handoff message
gt handoff crew # Hand off crew session
gt handoff mayor # Hand off mayor session
Note: Any molecule on the hook will be auto-continued by the new session. The SessionStart hook runs gt prime to restore context.
gt hook
Show or attach work on your hook.
The hook is the "durability primitive" - work on your hook survives session restarts, context compaction, and handoffs. When you restart (via gt handoff), your SessionStart hook finds the attached work and you continue from where you left off.
Syntax:
gt hook [bead-id] [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| (none) | Show what's on my hook |
| status [target] | Show what's slung on your hook |
| show <agent> | Show what's on an agent's hook (compact format) |
Flags:
| Flag | Description |
|------|-------------|
| -s, --subject string | Subject for handoff mail (optional) |
| -m, --message string | Message for handoff mail (optional) |
| -n, --dry-run | Show what would be done |
| -f, --force | Replace existing incomplete hooked bead |
| --json | Output as JSON (for status) |
Examples:
gt hook # Show what's on my hook
gt hook status # Same as above
gt hook gt-abc # Attach issue gt-abc to your hook
gt hook gt-abc -s "Fix the bug" # With subject for handoff mail
gt hook show greenplace/polecats/nux # What's nux working on?
gt hook show mayor # What's the mayor working on?
Related commands:
gt sling <bead>- Hook + start now (keep context)gt handoff <bead>- Hook + restart (fresh context)gt unsling- Remove work from hook
Diagnostic Commands
gt seance
Talk to predecessor sessions. Resume and query previous agent sessions.
Syntax:
gt seance [flags]
Flags:
| Flag | Description |
|------|-------------|
| --role string | Filter by role (crew, polecat, witness, etc.) |
| --rig string | Filter by rig name |
| -n, --recent int | Number of recent sessions to show (default: 20) |
| -t, --talk string | Session ID to commune with |
| -p, --prompt string | One-shot prompt (with --talk) |
| --json | Output as JSON |
Examples:
gt seance # List recent sessions
gt seance --role crew # Filter by role
gt seance --rig gastown # Filter by rig
gt seance --recent 10 # Last 10 sessions
gt seance --talk abc123 # Talk to predecessor
gt seance --talk abc123 -p "Where is X?" # One-shot question
Common Use Cases:
- Understanding predecessor decisions
- Finding where work was left off
- Debugging issues from previous sessions
gt feed
Show real-time activity feed from beads and gt events.
Syntax:
gt feed [flags]
Flags:
| Flag | Description |
|------|-------------|
| -f, --follow | Stream events in real-time |
| --no-follow | Show events once and exit |
| -n, --limit int | Maximum number of events (default: 100) |
| --since string | Show events since duration (e.g., 5m, 1h) |
| --mol string | Filter by molecule/issue ID prefix |
| --type string | Filter by event type |
| --rig string | Run from specific rig's beads directory |
| -w, --window | Open in dedicated tmux window |
| --plain | Use plain text output instead of TUI |
Examples:
gt feed # Launch TUI dashboard
gt feed --plain # Plain text output
gt feed --window # Open in dedicated tmux window
gt feed --since 1h # Events from last hour
gt feed --rig gastown # Use gastown rig's beads
gt feed --follow # Real-time stream
Common Use Cases:
- Monitoring agent activity in real-time
- Watching merge queue progress
- Debugging issue state changes
gt unsling
Remove work from an agent's hook (inverse of sling).
Syntax:
gt unsling [bead-id] [target]
Aliases: unhook
Flags:
| Flag | Description |
|------|-------------|
| -n, --dry-run | Show what would be done |
| -f, --force | Unsling even if work is incomplete |
Arguments:
- With no arguments: clears your own hook
- With bead ID only: only unslings if that specific bead is hooked
- With target only: operates on another agent's hook
- With both: unslings specific bead from specific agent
Examples:
gt unsling # Clear my hook
gt unsling gt-abc # Only unsling if gt-abc is hooked
gt unsling gastown/joe # Clear joe's hook
gt unsling gt-abc gastown/joe # Unsling gt-abc from joe
gt unsling --dry-run # Preview action
gt unsling --force # Unsling incomplete work
Common Use Cases:
- Reassigning work from one agent to another
- Clearing stuck hooks
- Cleaning up after aborted work
gt status
Show overall Gas Town status.
Syntax:
gt status [flags]
Flags:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
| --watch | Continuous status updates |
| --rig string | Focus on specific rig |
Examples:
gt status # Full status
gt status --json # JSON output
gt status --rig gastown # Focus on rig
gt status --watch # Live updates
gt prime
Output role context for current directory (used in agent initialization).
Syntax:
gt prime
Role Detection:
- Town root, mayor/, or
<rig>/mayor/-> Mayor context <rig>/witness/rig/-> Witness context<rig>/refinery/rig/-> Refinery context<rig>/polecats/<name>/-> Polecat context<rig>/crew/<name>/-> Crew context- deacon/ -> Deacon context
Examples:
gt prime # Output role context
Common Use Cases:
- Agent initialization on startup
- Role detection for scripts
- Context injection for Claude Code
gt whoami
Show current identity for mail commands.
Syntax:
gt whoami
Identity Detection:
- GT_ROLE env var (if set) - indicates an agent session
- No GT_ROLE - you are the overseer (human)
Examples:
gt whoami # Show current identity
Output includes:
- Identity string
- Source of identity (GT_ROLE, cwd, etc.)
- Additional env vars (GT_RIG, GT_POLECAT, GT_CREW)
gt version
Print version information.
Syntax:
gt version
Output:
gt version 0.1.1 (dev: main@abc123def)
gt witness
Manage witness (polecat supervisor) agents.
Syntax:
gt witness [subcommand] [rig] [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| start <rig> | Start the witness for a rig |
| stop <rig> | Stop the witness for a rig |
| attach <rig> | Attach to witness session |
| restart <rig> | Restart the witness |
| status [rig] | Show witness status |
Flags:
| Flag | Description |
|------|-------------|
| --json | Output as JSON |
Examples:
gt witness status # Show all witness status
gt witness status myrig # Show witness for specific rig
gt witness start myrig # Start witness for rig
gt witness stop myrig # Stop witness for rig
gt witness attach myrig # Attach to witness session
gt witness restart myrig # Restart witness
gt done
Signal work ready for merge queue. A convenience command for polecats that:
- Submits the current branch to the merge queue
- Auto-detects issue ID from branch name
- Notifies the Witness with the exit outcome
Exit types:
| Exit Type | Description |
|-----------|-------------|
| COMPLETED | Work done, MR submitted (default) |
| ESCALATED | Hit blocker, needs human intervention |
| DEFERRED | Work paused, issue still open |
Syntax:
gt done [flags]
Flags:
| Flag | Description |
|------|-------------|
| --issue string | Source issue ID (default: parse from branch name) |
| -p, --priority int | Override priority (0-4, default: inherit from issue) |
| --exit string | Exit type: COMPLETED, ESCALATED, or DEFERRED |
Examples:
gt done # Submit branch, notify COMPLETED
gt done --issue gt-abc # Explicit issue ID
gt done --exit ESCALATED # Signal blocker, skip MR
gt done --exit DEFERRED # Pause work, skip MR
gt done --priority 1 # Override priority
Note: This command requires the branch to be pushed. Run git push -u origin <branch> first.
Environment Variables
| Variable | Description |
|----------|-------------|
| GT_ROLE | Agent role (mayor, witness, polecat, crew, etc.) |
| GT_RIG | Current rig name |
| GT_POLECAT | Polecat name (for polecat role) |
| GT_CREW | Crew worker name (for crew role) |
| BD_ACTOR | Actor identity for beads operations |
Session Management
Gas Town uses tmux sessions for agent management:
| Session Pattern | Description |
|-----------------|-------------|
| gt-mayor | Mayor session |
| gt-deacon | Deacon session |
| gt-<rig>-witness | Witness for rig |
| gt-<rig>-refinery | Refinery for rig |
| gt-<rig>-polecat-<name> | Polecat session |
| gt-<rig>-crew-<name> | Crew session |
Quick Reference
Starting Work
gt up # Start all services
gt status # Check everything is running
gt crew start max # Start your crew worker
Dispatching Work
gt sling crew/max gt-issue-123 # Assign work
gt convoy create -i # Create parallel convoy
gt broadcast "Check mail" # Notify all workers
Monitoring
gt status # Overall status
gt polecat list # List workers
gt mq list # Check merge queue
gt mail inbox # Check messages
Cleanup
gt polecat gc # Clean up idle polecats
gt down # Shutdown infrastructure
gt stop --all # Emergency stop all polecats
Infrastructure Commands
gt daemon
Manage the Gas Town background daemon. The daemon is a simple Go process that pokes agents periodically and processes lifecycle requests.
Syntax:
gt daemon <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| start | Start the daemon |
| stop | Stop the daemon |
| status | Show daemon status |
| logs | View daemon logs |
Examples:
gt daemon start # Start daemon
gt daemon stop # Stop daemon
gt daemon status # Check daemon status
gt daemon logs # View daemon logs
gt deacon
Manage the Deacon - the hierarchical health-check orchestrator.
Syntax:
gt deacon <subcommand> [flags]
Aliases: dea
Subcommands:
| Subcommand | Description |
|------------|-------------|
| start | Start the Deacon session |
| stop | Stop the Deacon session |
| attach | Attach to the Deacon session |
| restart | Restart the Deacon session |
| status | Check Deacon session status |
| health-check <agent> | Send a health check ping to an agent |
| health-state | Show health check state for all agents |
| zombie-scan | Scan for idle polecats that should have been nuked |
| force-kill <session> | Force-kill an unresponsive agent session |
Examples:
gt deacon start # Start Deacon
gt deacon stop # Stop Deacon
gt deacon attach # Attach to Deacon session
gt deacon status # Check Deacon status
gt deacon health-state # Show health of all agents
gt deacon zombie-scan # Find orphaned polecats
gt mayor
Manage the Mayor - the global coordinator for Gas Town.
Syntax:
gt mayor <subcommand> [flags]
Aliases: may
Subcommands:
| Subcommand | Description |
|------------|-------------|
| start | Start the Mayor session |
| stop | Stop the Mayor session |
| attach | Attach to the Mayor session |
| restart | Restart the Mayor session |
| status | Check Mayor session status |
Examples:
gt mayor start # Start Mayor
gt mayor stop # Stop Mayor
gt mayor attach # Attach to Mayor session
gt mayor restart # Restart Mayor
gt mayor status # Check Mayor status
gt boot
Manage Boot - the Deacon's watchdog for triage.
Boot is a special dog that runs fresh on each daemon tick. It observes the system state and decides whether to start/wake/nudge/interrupt the Deacon.
Syntax:
gt boot <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| spawn | Spawn Boot for triage |
| status | Show Boot status |
| triage | Run triage directly (degraded mode) |
Location: ~/gt/deacon/dogs/boot/
Session: gt-deacon-boot
Recovery Commands
gt orphans
Find orphaned commits that were never merged to main.
Polecat work can get lost when sessions are killed before merge, Refinery fails to process, or network issues occur during push.
Syntax:
gt orphans [flags]
Flags:
| Flag | Description |
|------|-------------|
| --days int | Show orphans from last N days (default 7) |
| --all | Show all orphans (no date filter) |
Examples:
gt orphans # Last 7 days (default)
gt orphans --days=14 # Last 2 weeks
gt orphans --all # Show all orphans
gt checkpoint
Manage session checkpoints for crash recovery.
Checkpoints capture the current work state so that if a session crashes, the next session can resume from where it left off.
Syntax:
gt checkpoint <subcommand> [flags]
Subcommands:
| Subcommand | Description |
|------------|-------------|
| read | Read and display the current checkpoint |
| write | Write a checkpoint of current session state |
| clear | Clear the checkpoint file |
Checkpoint data includes:
- Current molecule and step
- Hooked bead
- Modified files list
- Git branch and last commit
- Timestamp
gt release
Release stuck in_progress issues back to pending.
Syntax:
gt release [flags]
Use when issues are stuck in in_progress state but no agent is working on them.
gt park
Park current work on a gate for async resumption.
When you need to wait for an external condition (timer, CI, human approval), park your work on a gate. When the gate closes, you'll receive wake mail.
Syntax:
gt park <gate-id> [flags]
Flags:
| Flag | Description |
|------|-------------|
| -m, --message string | Context notes for resumption |
| -n, --dry-run | Show what would be done |
Examples:
# Create a timer gate and park work on it
bd gate create --await timer:30m --title "Coffee break"
gt park <gate-id> -m "Taking a break, will resume auth work"
# Park on a human approval gate
bd gate create --await human:deploy-approval
gt park <gate-id> -m "Deploy staged, awaiting approval"
gt resume
Resume work that was parked on a gate, or check for handoff messages.
Syntax:
gt resume [flags]
Flags:
| Flag | Description |
|------|-------------|
| --status | Just show parked work status |
| --handoff | Check inbox for handoff messages instead |
| --json | Output as JSON |
Examples:
gt resume # Check for and resume parked work
gt resume --status # Just show parked work status
gt resume --handoff # Check inbox for handoff messages
Du fichier au premier essai
Le téléchargement contient exactement le contenu affiché sur cette fiche. Cette copie peut différer de la dernière version du dépôt source.
Consulter la source et ses fichiers annexesCe téléchargement contient uniquement SKILL.md. Les scripts, références, modèles et dépendances éventuels ne sont pas inclus. Récupérez les fichiers nécessaires depuis la source en conservant leur arborescence avant de poursuivre.
1. Télécharger le fichier
Télécharger SKILL.mdDepuis la racine d’un projet de test, enregistrez le fichier sous .skill-downloads/gastown-cli-commands-reference-mm5t8chm/SKILL.md, ou utilisez la commande suivante. Ce dossier sert à examiner la skill avant son installation.
Commande pour macOS/Linux ou WSL : un shell POSIX et curl sont nécessaires. Sous Windows sans WSL, utilisez le bouton de téléchargement et créez le dossier manuellement.
mkdir -p '.skill-downloads/gastown-cli-commands-reference-mm5t8chm' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/gastown-cli-commands-reference-mm5t8chm/download' --output '.skill-downloads/gastown-cli-commands-reference-mm5t8chm/SKILL.md.part' && mv '.skill-downloads/gastown-cli-commands-reference-mm5t8chm/SKILL.md.part' '.skill-downloads/gastown-cli-commands-reference-mm5t8chm/SKILL.md'2. Vérifier les prérequis
Lisez SKILL.md avant de l’utiliser. Si le fichier cite des scripts ou services externes, examinez-les et préparez les accès nécessaires dans votre environnement de test.
Extrait original : Quick Reference
Starting Work
gt up # Start all services
gt status # Check everything is running
gt crew start max # Start your crew worker
Dispatching Work
gt sling crew/max gt-issue-123 # Assign work
gt convoy create -i # Create parallel convoy
gt broadcast "Check mail" # Notify all workers
Monitoring
gt status # Overall status
gt polecat list # List workers
gt mq list # Check merge queue
gt mail inbox # Check messages
Cleanup
gt polecat gc # Clean up idle polecats
gt down # Shutdown infrastructure
gt stop --all # Emergency stop all polecats
Les plateformes de la fiche sont des indications du catalogue. Les chemins ci-dessous suivent la documentation des outils ; ils ne prouvent pas que cette skill y a été exécutée. Aucune version minimale de ces outils n’est attestée pour cette fiche.
3. Charger la skill
Le fichier ne fournit pas un en-tête name/description exploitable pour cette procédure native. Le téléchargement reste disponible pour lecture ; utilisez le mode manuel ci-dessous.
Mode manuel ou autre outil : joignez SKILL.md à une conversation, ou ouvrez-le dans un agent capable de lire les fichiers. Ajoutez les ressources annexes nécessaires. Il n’existe pas de dossier universel reconnu par tous les outils.
4. Faire un premier essai
Utilisez une entrée courte et fictive : un endpoint d’API, un petit document ou un brief d’interface. Dans une nouvelle conversation, joignez le fichier ou donnez son chemin à l’agent, puis adaptez cette demande :
Lis .skill-downloads/gastown-cli-commands-reference-mm5t8chm/SKILL.md et applique ses instructions à cet exemple : [décrire une entrée de test]. Indique les fichiers, dépendances ou accès qui manquent. Produis le résultat demandé par la skill sans inventer de données ni exécuter de script non examiné.Résultat attendu : une sortie correspondant aux instructions de la skill, ou une liste précise des prérequis manquants. Une simple confirmation de lecture ne valide pas le premier usage. Si une ressource externe est indisponible, ce parcours reste à tester.
Configurer un canal de discussion avec SecretRefs
DevOps
Configurez des canaux de discussion avec des commandes non interactives, les secrets étant uniquement des SecretRefs. Couvre les étapes de collecte, modification, réparation et vérification.
Enquêter sur les échecs CI de ClickHouse
DevOps
Investigue un échec CI de ClickHouse de bout en bout à partir d'une URL de PR ou de rapport S3. Récupère les tests échoués, classe flaky vs vraie régression, et cherche des issues/fixes existants.
Revue de PR Ansible
DevOps
Examine une PR Ansible selon le processus standardisé du projet, incluant la validation des changelogs, la couverture de tests et le suivi des commentaires.