generated: '2026-08-24'
method: searched
source: https://www.agentsea.dev/docs/cli/
docs: https://www.agentsea.dev/docs/cli/
name: AgentSea CLI
binary: sea
package: '@lov3kaizen/agentsea-cli'
version: '1.2.1'
published: '2026-07-02'
license: MIT
note: >-
  First-party interactive CLI for the AgentSea ADK, published to npm on the same lockstep version
  line as the rest of the @lov3kaizen/agentsea-* suite. It manages local configuration, agents,
  providers and Ollama models, and hosts the agentic coding session (`sea code`) added in v0.6.0.
  It is a local developer tool - it does not call a vendor-operated AgentSea API.
install:
- method: npm
  command: npm install -g '@lov3kaizen/agentsea-cli'
- method: pnpm
  command: pnpm add -g '@lov3kaizen/agentsea-cli'
- method: yarn
  command: yarn global add '@lov3kaizen/agentsea-cli'
requirements:
  node: '>=20.0.0'
config_paths:
  macos: ~/Library/Preferences/agentsea-cli
  linux: ~/.config/agentsea-cli
  windows: '%APPDATA%/agentsea-cli'
  format: JSON (providers, agents, defaultAgent, voice)
commands:
- group: setup
  command: sea init
  description: Interactive setup wizard - choose provider, enter API key, select default model, configure voice providers, save configuration.
- group: chat
  command: sea chat
  description: Interactive chat session with the default or a named agent.
  flags:
  - --agent <name>
  - --voice
  - --model <model>
- group: coding
  command: sea code
  description: >-
    Interactive agentic coding session with 13 built-in coding tools (file_read, file_write,
    file_list, code_edit, glob, grep, shell_execute, git_status and more). Works with any
    configured provider, cloud or local.
  flags:
  - --provider <provider>
  - --model <model>
  - --verbose
  - --maxIterations <n>
- group: agent
  command: sea agent create
  description: Create an agent with provider, model, system prompt, temperature, max tokens and optional voice STT/TTS.
  flags:
  - --provider
  - --model
  - --system-prompt
  - --temperature
  - --max-tokens
  - --voice-stt
  - --voice-tts
- group: agent
  command: sea agent list
  description: List configured agents with provider, model and voice configuration.
- group: agent
  command: sea agent get <name>
  description: Show a single agent's configuration.
- group: agent
  command: sea agent run <name> "<prompt>"
  description: Run an agent once; supports --voice-input and --voice-output audio files.
- group: agent
  command: sea agent delete <name>
  description: Delete an agent.
- group: agent
  command: sea agent default <name>
  description: Set the default agent.
- group: model
  command: sea model pull <model>
  description: Pull a local model (Ollama).
- group: model
  command: sea model list
  description: List locally installed models with sizes.
- group: model
  command: sea model info <model>
  description: Show model size, parameters, quantization, family and license.
- group: model
  command: sea model remove <model>
  description: Remove a local model.
- group: provider
  command: sea provider add <provider> --api-key <key>
  description: Add and configure an LLM provider.
- group: provider
  command: sea provider list
  description: List configured providers with masked keys and defaults.
- group: provider
  command: sea provider test <provider>
  description: Test connectivity to a configured provider.
- group: provider
  command: sea provider remove <provider>
  description: Remove a provider configuration.
- group: meta
  command: sea --version
  description: Print the installed CLI version.
key_flows:
- name: Fully local, zero-cost setup
  steps:
  - Install Ollama
  - sea init (choose Ollama)
  - sea model pull llama3.2
  - sea chat
- name: Voice-enabled assistant
  steps:
  - sea agent create voice-bot --voice-stt openai-whisper --voice-tts openai-tts
  - sea chat --agent voice-bot --voice
  - sea agent run voice-bot --voice-input question.mp3 --voice-output answer.mp3
- name: Multi-agent workflow
  steps:
  - sea agent create researcher / writer / coder with different models
  - sea agent run <name> "<task>"