cli
已验证 · 实测可装 AtlasCloudAI
功能简介
AtlasCloud CLI安装器和发布产物
可用 — 实测通过,早期项目
AtlasCloud CLI安装器和发布产物 实测能干净安装、正常启动。早期项目,但功能可用。
「已验证」表示我们的自动化 CI 在干净 profile 里实际执行了 dsh plugin add 并启动成功——仅此而已。功能描述与版本兼容性均为作者声明。这不是安全审计,也不代表对第三方代码的背书。
README
Atlas Cloud CLI
Call Atlas Cloud LLM, image, and video APIs from your shell, scripts, and CI jobs — with image/video/audio inputs for multimodal chat.
→ Get your free Atlas Cloud API key — 300+ models, one key, OpenAI-compatible.
This repository hosts public installers, release artifacts, and lightweight package-manager wrappers for the atlas CLI. The Go source repository is maintained separately.
Supported Models
🎬 Video (174) — Seedance 2.5 · MiniMax H3 · Youchuan V8.2 · Seedance 2.0 Mini · HappyHorse-1.1 · Gemini Omni Flash
🎨 Image (118) — Grok Imagine Image 2.0 · Qwen Image 3.0 Pro · Seedream v5.0 Pro · Qwen Image 3.0
🧊 3D (7) — Seed3D 2.0 · Hunyuan 3D Rapid · Hunyuan 3D Pro · Tripo H3.1
💬 LLM (65) — Grok 4.6 · DeepSeek V4 Flash 0731 · Qwen3.8 Max · Kimi K3
📚 Explore more — all 398 live models »
🎬 Newest video models — Seedance 2.5 · Kling 4.0 · Wan 3.0 · Kling Video O3.
ℹ️ Audio: the CLI accepts audio as input for multimodal chat (
atlas chat --audio @file.mp3, e.g. transcription withqwen/qwen3-vl-8b-instruct). Audio generation (TTS / music) and dedicated 3D/audio subcommands are not in the CLI yet — track #3 or use the MCP server / REST API meanwhile.
Availability, parameters, and pricing vary by model. Use atlas models get andatlas generate cost against the live catalog before automating billable calls.
Contents
- Supported Models
- Install
- API caller workflows
- Guides
- Commands
- Global Flags
- Updating
- Uninstall
- Troubleshooting
- Support
- More Atlas Cloud Tools
- License
Install
macOS / Linux
curl -fsSL https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.sh | sh
Options:
INSTALLER=https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.sh
# Custom prefix, no sudo if the directory is writable
curl -fsSL "$INSTALLER" | sh -s -- --prefix="$HOME/.local"
The installer always uses the latest GitHub Release, downloads the matching archive, and verifies it against checksums.txt.
Windows
irm https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.ps1 | iex
Options:
# Custom install dir
$env:ATLAS_INSTALL_DIR="$HOME\bin"; irm https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.ps1 | iex
# Skip adding atlas.exe to user PATH
$env:ATLAS_NO_PATH="1"; irm https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.ps1 | iex
The Windows installer always uses the latest GitHub Release, downloads the matching windows_amd64 or windows_arm64 zip, verifies it against checksums.txt, installs atlas.exe, and adds the install directory to the user PATH by default.
Homebrew
brew install AtlasCloudAI/tap/atlascloud
The formula is named atlascloud, but it installs the atlas command.
npm
npm install -g atlascloud-cli
The npm package is a thin wrapper. Its postinstall script downloads the matching prebuilt release archive and verifies the checksum before exposing atlas.
Manual
Download the archive for your OS and architecture from Releases, extract it, and place the binaries in your PATH.
API caller workflows
Authenticate once, then choose the command path that matches the API call you
want to make. Discovery and cost commands are safe to run before starting a
billable generation.
atlas auth login # interactive
atlas auth login --token "$ATLASCLOUD_API_KEY" # CI / non-interactive
atlas auth status
| Job | Command pattern | Notes |
|---|---|---|
| Discover available models | atlas models list --type video --json |
Replace video with chat or image; avoid hard-coding stale model IDs. |
| Inspect model parameters | atlas models get MODEL_ID --json |
Read required fields, defaults, and vendor-specific parameter names. |
| Estimate generation cost | atlas generate cost video MODEL_ID ... --json |
Use cost image or cost video; calls the pricing endpoint only. |
| Call a chat or multimodal model | atlas chat --model MODEL_ID "prompt" |
Supports text plus --image, --video, and --audio for capable models. |
| Start an image/video job | atlas generate image MODEL_ID ... |
Use generate image or generate video; add --no-wait --json for async scripts. |
| Continue an async job | atlas generate get PREDICTION_ID / atlas generate wait PREDICTION_ID |
Poll status or wait until completion. |
| Script and CI usage | atlas --json ... | jq ... |
Non-TTY output is JSON by default; --json makes it explicit. |
Explore without model calls
atlas models list --type video --json | jq -r '.models[].id'
atlas models search seedance --type video --json
atlas models get bytedance/seedance-2.5/text-to-video --json
Estimate cost before generation
atlas generate cost video bytedance/seedance-2.5/text-to-video \
-p "A product shot slowly rotates on a clean white background" \
--duration 5 \
--resolution 720p \
--param generate_audio=false \
--json
Make API calls
atlas chat --model deepseek-ai/DeepSeek-V3-0324 "Return only a JSON object with status=ok"
PRED=$(atlas generate image google/nano-banana-2/text-to-image \
-p "minimal product photo on a white background" \
--no-wait --json | jq -r '.id')
atlas generate wait "$PRED"
Prefer environment variables? Copy .env.example to .env and set ATLASCLOUD_API_KEY.
More API caller scripts (discovery-first call → cost-aware generation → scripted pipeline → CI JSON job) live in examples/.
Guides
- API caller workflows — how to use Atlas CLI as a shell-first API client in local scripts, CI jobs, and backend automation.
Commands
| Command | Purpose |
|---|---|
atlas auth |
Log in, log out, inspect local auth state |
atlas chat |
Send a chat completion request |
atlas models |
List and inspect available models |
atlas generate |
Generate images and videos, poll job status |
atlas account |
Manage account selection |
atlas version |
Print build information |
Run atlas --help or atlas <command> --help for full flag reference.
Global Flags
| Flag | Purpose |
|---|---|
--json |
Force machine-readable JSON output |
--no-color |
Disable ANSI color |
--quiet |
Suppress spinners and progress text |
--verbose |
Print debug output to stderr |
Updating
# curl installer
curl -fsSL https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.sh | sh
# Homebrew
brew update && brew upgrade atlascloud
# npm
npm install -g atlascloud-cli@latest
Uninstall
# curl installer, default prefix
sudo rm -f /usr/local/bin/atlas
# Homebrew
brew uninstall atlascloud
# npm
npm uninstall -g atlascloud-cli
Troubleshooting
Not logged in — run atlas auth login.
Unknown model — run atlas models list or atlas models search <keyword>.
Installer checksum failure — do not run the downloaded archive. Retry the install or open an issue with the exact URL and version.
Support
Bugs and feature requests: GitHub Issues. Please include atlas version, your OS/arch, install method, and the exact command that failed.
More Atlas Cloud Tools
- 🧰 Want to use it from the terminal? → atlascloud-cli
- 🤖 Want to use it in Claude Code / Cursor? → Install the Atlas Cloud MCP Server
- 🎬 Want it as a Claude Code / Codex / Gemini CLI Skill? → Install atlas-cloud-skills
- 🎨 ComfyUI nodes → atlascloud_comfyui
- 🔁 n8n nodes → n8n-nodes-atlascloud
- 💬 Join our Discord → discord.gg/MWmMr4q9es
- 🌐 Website → atlascloud.ai
License
MIT
安装
装一次目录插件,之后本站所有插件都能让 DeepSeek Harness 自动找、自动装:
dsh plugin add dshbase-catalog 然后对 agent 说「帮我装 cli」,它会在目录里找到并自动安装。文档:dshbase-catalog · 已验证场景包。
该插件是 GitHub 源码(未发 npm)——直接从仓库装:
Web profile:
dsh plugin --profile web add github:AtlasCloudAI/cli Headless(CLI)profile:
dsh plugin --profile headless add github:AtlasCloudAI/cli 实测报告
验证通过:从 GitHub 源码完成 L1 安装 + L2 加载 + L3 运行(dsh 0.1.0-rc.6)。
使用场景
改善终端体验——TUI、命令面板或 shell 集成——让 dsh 在终端里如原生一般。
适合谁
终端优先、从命令行驱动 dsh、想让它像熟悉的 CLI agent 一样工作的开发者。
二次开发建议
TUI 布局和 shell 集成是缝——加面板、快捷键或新的 shell 钩子。