suggest_tags¶
Suggest NovelAI prompt tags that complete or refine a partial prompt. The suggestions are ranked by NovelAI’s training-set tagger and returned with their training occurrence count.
Parameters¶
Name |
Type |
Default |
Notes |
|---|---|---|---|
|
|
required |
Partial prompt text to complete. |
|
|
|
Tagger vocabulary (matches a |
|
|
|
ISO 639-1 response language (e.g. |
Returns¶
list[dict] — list of tag descriptors, each with:
Key |
Type |
Notes |
|---|---|---|
|
|
Human-readable description (localized to |
|
|
The tag string itself (use this in your prompt). |
|
|
Training-set occurrence count (higher = more reliable). |
Example¶
result = await ctx.session.call_tool("suggest_tags", {
"prompt": "1girl, fox ears,",
"model": "nai-diffusion-4-5-full",
"language": "en",
})
# result: [
# {"description": "fox tail", "text": "fox tail", "count": 12450},
# {"description": "autumn leaves", "text": "autumn leaves", "count": 8230},
# ...
# ]
Tips¶
Tip
Iterative prompt building: ask an agent to call suggest_tags after
each prompt edit, then accept the top suggestion that fits the scene. This
is a fast way to discover tags you wouldn’t have thought of.
Tag count vs quality
count measures how often the tag appears in NovelAI’s training data — not
how often it’s useful for your prompt. A rare tag (low count) can still
be the perfect descriptor for your scene.