Multimodal AI
Image Captioning API MCP
Extract descriptive prompts or selling-point text from one or more images. Call it via REST API or MCP, or try it in the online playground.
Image Captioning API AI Prompts & Use Cases
After connecting this MCP tool to your AI assistant, you can ask the questions below. Replace bracketed placeholders with your own inputs and use the supported values in Arguments.
New to Nexscope MCP? Follow the connection guide, then use the prompts below. The Integration and Parameters & Response tabs contain the connection steps, call example, and technical reference.
1. Describe A Similar Image
Prompt: Write a prompt for a visually similar image from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
2. Describe A Scene
Prompt: Write a scene prompt from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
3. Plan Image Animation
Prompt: Analyze the images for an image-to-video brief from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
4. Describe Apparel Selling Points
Prompt: Describe visible apparel selling points from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
5. Describe Product Selling Points
Prompt: Describe visible product selling points from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
6. Draft A Sales Video Script
Prompt: Draft a sales-video script grounded in the visible product from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
7. Plan A Recreated Video
Prompt: Create a video-recreation brief using the reference video from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials. Use [reference video], [target country], [target language] and [duration in seconds] for the recreation brief.
8. Choose An Image Model
Prompt: Write a prompt for a visually similar image from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials. Use [supported image or analysis model] for this task, keeping the same references and creative brief.
What You Need to Provide
Provide assets you may use and that meet this tool’s documented file, dimension and count limits.
A pending task is not a completed image, video or text result. Retrieve the existing task rather than submitting a duplicate.
Step 1 · Connect
Connect Your AI Assistant
Connect a compatible MCP client to Nexscope, then select this tool from your assistant. Use the connection settings in your client to complete authentication.
POST https://api.nexscope.ai/mcp
MCP-Protocol-Version: 2025-11-25
Content-Type: application/jsonAuthentication
Use OAuth 2.1 Authorization Code with PKCE and the mcp:tools scope. API keys remain supported as a compatibility option.
Authorization: Bearer nk-xxxxxxxxxxxxxxxxxUsing an API key? Get your key from API Access and include it in the Authorization header shown above.
Call this MCP tool
nexscope_image_captioninghttps://api.nexscope.ai/mcptools/callInitialize an MCP 2025-11-25 Streamable HTTP session at https://api.nexscope.ai/mcp, then use tools/call.
{
"jsonrpc": "2.0",
"id": "image-captioning-1",
"method": "tools/call",
"params": {
"name": "nexscope_image_captioning",
"arguments": {
"imageUrlList": [
"https://example.com/product.jpg"
],
"keyword": "phone case",
"provider": "API_PRE_POINT",
"type": 71
}
}
}MCP 2025-11-25 · Streamable HTTPProgress is delivered on the original POST SSE response. GET SSE carries notifications; reconnecting must query existing asynchronous work rather than submit again.
Asynchronous task workflow
- 1. Create once
Keep the returned taskId.
- 2. Query with backoff
Use nexscope_media_task_result after 5 seconds, then 10, 20, and 30 seconds.
- 3. Use final result
Stop at success, failure, or timeout.
{
"jsonrpc": "2.0",
"id": "image-captioning-result-1",
"method": "tools/call",
"params": {
"name": "nexscope_media_task_result",
"arguments": {
"skillSlug": "image-captioning",
"taskId": "TASK_ID_FROM_THE_GENERATION_RESULT"
}
}
}Arguments
| Name | Type | Required | Description |
|---|---|---|---|
type | integer | required | Analysis type. Available options
Example: 71 |
imageUrlList | array | required | Public image URLs used for analysis. Provide no more than 6 images for modes that accept multiple images. Example: [
"https://example.com/product.jpg"
] |
keyword | string | optional | Optional keyword context. Example: phone case |
provider | string | optional | Analysis provider. Available options
Example: API_PRE_POINT |
refVideoUrl | string | optional | Reference video URL. Required when type is 89. Example: https://example.com/reference-video.mp4 |
saleCountry | string | optional | Target sale country code. Example: US |
targetLanguage | string | optional | Target language for generated text. Used by video-reclone analysis. Example: English |
duration | integer | optional | Target duration in seconds. Used by video-reclone analysis. Example: 15 |
Response fields
Treat structuredContent as the programmatic API response payload.
| Name | Type | Required | Description |
|---|---|---|---|
code | integer | required | 0 means accepted or successful; nonzero is a platform error. |
cost | string | required | Elapsed milliseconds, never credits; -1 outside web requests. |
data | any | required | Data value used by this API operation. |
msg | string | null | required | Msg value used by this API operation. |
time | string | required | Server local time: yyyy-MM-dd HH:mm:ss. |
traceId | string | null | required | Trace id value used by this API operation. |
ts | string | required | Epoch milliseconds. |
Any of option 1 — requires taskId, status, result
| Name | Type | Required | Description |
|---|---|---|---|
data | object | required | Data value used by this API operation. |
data.result | any | required | Completed generation output retained from the task result. Fields vary by capability; generated resources use Nexscope-hosted URLs and any count represents charged Nexscope credits. This is the business output, not a task creation receipt. |
data.status | string | required | Status value used by this API operation. Allowed: PENDING, RUNNING, SUCCEEDED, FAILED, TIMEOUT |
data.taskId | string | required | Task id value used by this API operation. |
Any of option 1
| Name | Type | Required | Description |
|---|---|---|---|
data.result | object | required | Completed generation output retained from the task result. Fields vary by capability; generated resources use Nexscope-hosted URLs and any count represents charged Nexscope credits. This is the business output, not a task creation receipt. |
Any of option 2
| Name | Type | Required | Description |
|---|---|---|---|
data.result | null | required | Completed generation output retained from the task result. Fields vary by capability; generated resources use Nexscope-hosted URLs and any count represents charged Nexscope credits. This is the business output, not a task creation receipt. |
Any of option 2
| Name | Type | Required | Description |
|---|---|---|---|
data | null | required | Data value used by this API operation. |
Format Illustration
Illustrative example with fictional data, not a live tool call.
Illustrative Question
Write a prompt for a visually similar image from [reference images], using [keyword context]. Do not invent specifications that cannot be supported by the supplied materials.
Illustrative Response Explanation
The illustrative request has been accepted as task example-task and is PENDING. No completed output is available yet. Keep this task ID and check its result; do not submit the same request again just because processing has not finished.
About this sample: Illustrative format example with fictional values, not a live tool call. Replace example inputs with your own permitted assets or public identifiers.
Published sample data is not available for this page. A format illustration or workflow guide is not a substitute for a published result.
View the MCP Request & Response Format
Examples
Illustrative example with fictional data, not a live tool call.
The JSON-RPC envelope below illustrates how the shown arguments and payload fit into MCP. It is not captured network traffic or proof of a successful call. Workflow-only examples do not include a result.
MCP Request
{
"jsonrpc": "2.0",
"id": "image-captioning-example",
"method": "tools/call",
"params": {
"name": "nexscope_image_captioning",
"arguments": {
"imageUrlList": [
"https://example.com/product.jpg"
],
"keyword": "phone case",
"provider": "API_PRE_POINT",
"type": 71
}
}
}MCP Response Format
{
"jsonrpc": "2.0",
"id": "image-captioning-example",
"result": {
"content": [
{
"type": "text",
"text": "{\"code\":0,\"cost\":\"-1\",\"data\":{\"taskId\":\"example-task\",\"status\":\"PENDING\",\"result\":null},\"msg\":null,\"time\":\"2026-01-01 00:00:00\",\"traceId\":null,\"ts\":\"0\"}"
}
],
"structuredContent": {
"code": 0,
"cost": "-1",
"data": {
"taskId": "example-task",
"status": "PENDING",
"result": null
},
"msg": null,
"time": "2026-01-01 00:00:00",
"traceId": null,
"ts": "0"
},
"isError": false
}
}