Overview
The Hive CLI (hive) provides commands for running agents, managing sessions, testing, and more.
Installation
hive command will be available in your PATH.
Global Options
Anthropic model to use for all commands
Command Categories
Agent Execution
hive run- Run an exported agenthive shell- Interactive agent sessionhive dispatch- Dispatch to multiple agentshive tui- Launch interactive TUI dashboard
Agent Information
hive info- Show agent informationhive validate- Validate an agenthive list- List available agents
Session Management
hive sessions list- List agent sessionshive sessions show- Show session detailshive sessions checkpoints- List checkpointshive pause- Pause running sessionhive resume- Resume from checkpoint
Testing
hive test-run- Run tests for an agenthive test-debug- Debug a failed testhive test-list- List testshive test-stats- Show test statistics
Development
hive code- Launch Hive Coder (agent builder)hive setup-credentials- Interactive credential setuphive serve- Start HTTP API server
Common Commands
info
Show agent information.Path to agent folder (containing agent.json)
Output as JSON
validate
Validate an exported agent.Path to agent folder
- Graph structure validation
- Tool availability check
- Warnings and suggestions
list
List available agents.Directory to search for agents
shell
Start an interactive agent session (REPL).Path to agent folder (optional, can select interactively)
Directory containing agents
Enable multi-agent mode with orchestrator
Disable human-in-the-loop approval (auto-approve all steps)
- Natural language input (auto-formatted with Haiku)
- JSON input support
- Session memory across inputs
- Special commands:
/info- Show agent details/nodes- Show agent nodes/reset- Reset conversation state/quit- Exit
dispatch
Dispatch request to multiple agents via orchestrator.Directory containing agent folders
Input context as JSON string
Description of what you want to accomplish
Specific agent names to use (default: all in directory)
Only output the final result JSON
code
Launch Hive Coder (interactive agent builder).LLM model to use (any LiteLLM-compatible name)
Session Management
sessions list
List agent sessions.Path to agent folder
Filter by status: all, active, failed, completed, paused
Show only sessions with checkpoints
sessions show
Show session details.Path to agent folder
Session ID to inspect
Output as JSON
sessions checkpoints
List session checkpoints.Path to agent folder
Session ID
pause
Request graceful pause of a running session.Path to agent folder
Session ID to pause
resume
Resume a paused or failed session from a checkpoint.Path to agent folder
Session ID to resume
Specific checkpoint ID to resume from (default: latest)
Resume in TUI dashboard mode
setup-credentials
Interactive credential setup.Path to agent folder (optional - runs general setup if not specified)
serve
Start HTTP API server.Host to bind
Port to listen on
Agent path(s) to preload (repeatable)
LLM model for preloaded agents
Open dashboard in browser after server starts
- REST + SSE APIs for agent control
- Web dashboard for monitoring
- Multi-agent support
Examples
Run an agent
Interactive session
Test an agent
Manage sessions
Environment Variables
Required for Claude models
Required for GPT models
Base path for agent storage
Logging level: DEBUG, INFO, WARNING, ERROR
Exit Codes
0: Success1: Error (validation failed, execution failed, etc.)