nodecode-dashscope-image
DashScope Image: pictures from Alibaba's Wan 2.7 models through a Qwen Cloud key, as the dashscope_image tool and /dashscope. A Nodecode rewrite of rriggs/hermes-plugin-dashscope-image by rriggs.
The original is an image backend for another agent, plugged into its one image_generate tool. Here the same wire is a cell of its own: one tool, one command, and the pictures written under a folder you name.
| Surface | What it does |
|---|---|
dashscope_image |
One picture from a prompt: subject, setting, style, light, framing. The answer names the file, its size in bytes and its pixel size. |
dashscope_image with images |
An edit or a restyle: up to 4 source pictures - local paths, https URLs or data URIs - and a prompt that says what to change. |
/dashscope |
The same from your side of the keyboard: /dashscope a red fox in snow, with --landscape, --square, --portrait, --model and --image PATH in front. Typed alone it asks for the prompt. |
Aspect picks the size the Wan models take, in DashScope's W*H form: landscape 1280*720, square 1024*1024, portrait 720*1280. A call that names a source image upgrades wan2.7-image to wan2.7-image-pro on its own, because the fast model is text-to-image only - the original plugin does the same.
Pictures are downloaded as soon as they are drawn, because the OSS links DashScope answers with expire after about a day.
Get it
Tell the agent "add the dashscope-image cell", or run nodecode add nc://dashscope-image.
Set it up
You need a Qwen Cloud (Alibaba Cloud Model Studio) key: a token plan key (sk-sp-*), a pay-as-you-go key (sk-ws-*), or a plain DashScope API key. Put it in an environment variable, then in /settings (or in the dashscope section of config.jsonc):
"dashscope": {
"api": "https://token-plan.ap-southeast-1.maas.aliyuncs.com",
"api_key_env": "QWEN_API_KEY",
"model": "wan2.7-image",
"output_dir": "~/.nodecode/images"
}
| Setting | What it does |
|---|---|
api |
The DashScope base. The token-plan host is the default; a pay-as-you-go key wants https://dashscope-intl.aliyuncs.com. HTTPS unless the host is loopback, LAN, Tailscale or .local: the key never crosses the public internet in clear text. |
api_key_env |
The name of the environment variable holding the key, never the key itself. Required. |
api_key_file |
A file holding the key instead, which the file's own permissions guard. |
model |
wan2.7-image (fast) or wan2.7-image-pro (higher fidelity, takes a source image). |
output_dir |
Where pictures are written, ~/.nodecode/images unless you say otherwise. |
timeout |
How long one request may take, in seconds. A render is slow. |
Use it
You do not call the tool; the model does. "Draw a wide shot of a red fox in snow", "restyle ~/pictures/fox.png with the sky at dusk", "make a portrait illustration of a heron in ink". A person can call the same function from eval, in the dashscope package:
(dashscope:generate-image "a red fox in snow" :aspect "landscape")
(help :dashscope-image) is the model's manual for the tool, kept while the cell runs.
What it sends, and what it will not
The prompt, the model, the size, and any source images go to the api base you set, and to nowhere else. Nothing else is read: no other app's tokens, no browser profile, no environment beyond the one variable you name. A local source image is sent base64-inlined, as the original does, because DashScope fetches a URL from its own side; it must really be a PNG, JPEG, WebP or GIF - checked by its first bytes, at most 10 MB - so a mistyped path fails before any request instead of uploading a text file. A picture the answer links is fetched back over https (or plain HTTP only from a local network).
A failed call says the HTTP status, DashScope's own code and message with your key blanked out, and the hint that usually fixes it: a wrong key, an api base belonging to another region or plan, a model the key cannot reach, or a rate limit.
