lovart

Imported from okou-ai/vm0-skills (lovart/SKILL.md). Install upstream with npx skills add okou-ai/vm0-skills --skill lovart. Copyright stays with the author.

Troubleshooting

If requests fail, run okou doctor check-connector --env-name LOVART_ACCESS_KEY or okou doctor check-connector --url https://lgw.lovart.ai/v1/openapi/mode/query --method POST

Authentication

All requests use AK/SK HMAC-SHA256 signature authentication. Each request requires five signed headers:

Environment Variables

Key Endpoints

Base URL: https://lgw.lovart.ai

1. Generate Design (Chat)

POST /v1/openapi/chat

Send a natural-language prompt to generate images, videos, or other visual content. Returns a thread_id for tracking.

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/chat\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/chat" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/chat" -d @/tmp/lovart_chat.json

Write to /tmp/lovart_chat.json:

{
  "prompt": "<your-design-description>",
  "project_id": "<your-project-id>"
}

To continue a conversation, add "thread_id": "<your-thread-id>".

To attach reference images, add "attachments": ["<cdn-url>"].

To specify a preferred model, add "tool_config": {"prefer_tool_categories": {"IMAGE": ["generate_image_flux_2_max"]}}.

To force a specific tool (e.g. upscale), add "tool_config": {"include_tools": ["upscale_image"]}.

Response: {"thread_id": "..."}

2. Check Generation Status

GET /v1/openapi/chat/status?thread_id=<thread-id>

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "GET\n/v1/openapi/chat/status\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s "https://lgw.lovart.ai/v1/openapi/chat/status?thread_id=<your-thread-id>" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: GET" --header "X-Signed-Path: /v1/openapi/chat/status"

Response includes status (running, done, pending_confirmation, failed) and progress (0-100).

3. Get Generation Result

GET /v1/openapi/chat/result?thread_id=<thread-id>

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "GET\n/v1/openapi/chat/result\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s "https://lgw.lovart.ai/v1/openapi/chat/result?thread_id=<your-thread-id>" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: GET" --header "X-Signed-Path: /v1/openapi/chat/result"

Response includes items with generated artifacts (image URLs, video URLs) and thread_id.

4. Confirm High-Cost Operation

POST /v1/openapi/chat/confirm

When status is pending_confirmation, confirm to proceed with high-cost operations (e.g. video generation).

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/chat/confirm\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/chat/confirm" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/chat/confirm" -d @/tmp/lovart_confirm.json

Write to /tmp/lovart_confirm.json:

{
  "thread_id": "<your-thread-id>"
}

5. Create Project

POST /v1/openapi/project/save

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/project/save\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/project/save" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/project/save" -d @/tmp/lovart_project.json

Write to /tmp/lovart_project.json:

{
  "project_id": "",
  "project_name": "<your-project-name>",
  "project_type": 3
}

Response includes project_id. Use the returned ID in subsequent chat requests.

6. Set Generation Mode

POST /v1/openapi/mode/set

Switch between fast mode (costs credits, no queue) and unlimited mode (free, may queue).

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/mode/set\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/mode/set" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/mode/set" -d @/tmp/lovart_mode.json

For fast mode, write to /tmp/lovart_mode.json:

{
  "unlimited": false
}

For unlimited mode, write to /tmp/lovart_mode.json:

{
  "unlimited": true
}

7. Query Generation Mode

POST /v1/openapi/mode/query

TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/mode/query\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/mode/query" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/mode/query" -d '{}'

Available Models

Use model names in tool_config.prefer_tool_categories or tool_config.include_tools:

IMAGE: generate_image_midjourney, generate_image_nano_banana_pro, generate_image_nano_banana_2, generate_image_nano_banana, generate_image_gpt_image_1_5, generate_image_seedream_v5, generate_image_seedream_v4_5, generate_image_seedream_v4, generate_image_imagen_v4, generate_image_flux_2_max, generate_image_flux_2_pro

VIDEO: generate_video_seedance_v2_0, generate_video_seedance_v2_0_fast, generate_video_seedance_pro_v1_5, generate_video_kling_v3, generate_video_kling_v3_omni, generate_video_kling_v2_6, generate_video_kling_omni_v1, generate_video_veo3_1, generate_video_veo3_1_fast, generate_video_veo3, generate_video_sora_v2_pro, generate_video_sora_v2, generate_video_wan_v2_6, generate_video_hailuo_v2_3, generate_video_vidu_q2

3D: generate_3d_tripo

UTILITY: upscale_image

Common Workflows

Generate an Image and Retrieve the Result

# Step 1: Send chat request
TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "POST\n/v1/openapi/chat\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s -X POST "https://lgw.lovart.ai/v1/openapi/chat" --header "Content-Type: application/json" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: POST" --header "X-Signed-Path: /v1/openapi/chat" -d '{"prompt": "A modern logo design with gradient blue to purple", "project_id": "<your-project-id>"}'

# Step 2: Poll status until done (replace <thread-id> from Step 1)
TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "GET\n/v1/openapi/chat/status\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s "https://lgw.lovart.ai/v1/openapi/chat/status?thread_id=<your-thread-id>" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: GET" --header "X-Signed-Path: /v1/openapi/chat/status"

# Step 3: Get result when status is "done"
TIMESTAMP=$(date +%s) && SIGNATURE=$(echo -ne "GET\n/v1/openapi/chat/result\n${TIMESTAMP}" | openssl dgst -sha256 -hmac "$LOVART_SECRET_KEY" | awk '{print $NF}') && curl -s "https://lgw.lovart.ai/v1/openapi/chat/result?thread_id=<your-thread-id>" --header "X-Access-Key: $LOVART_ACCESS_KEY" --header "X-Timestamp: ${TIMESTAMP}" --header "X-Signature: ${SIGNATURE}" --header "X-Signed-Method: GET" --header "X-Signed-Path: /v1/openapi/chat/result"

Upscale an Image

Use include_tools to force the upscale tool:

Write to /tmp/lovart_chat.json:

{
  "prompt": "upscale this image to 4K",
  "project_id": "<your-project-id>",
  "attachments": ["<your-image-cdn-url>"],
  "tool_config": {
    "include_tools": ["upscale_image"]
  }
}

Guidelines

1. Signature freshness: The HMAC signature includes a timestamp — recompute TIMESTAMP and SIGNATURE for every request, as the server rejects stale signatures
2. Signed path must match: The X-Signed-Path header must exactly match the request path (without query string), e.g. /v1/openapi/chat not /v1/openapi/chat?thread_id=xxx
3. Polling: Image generation typically takes 10-30s; video generation can take minutes — poll chat/status until status is done
4. Pending confirmation: When status is pending_confirmation, call chat/confirm before the generation proceeds — do NOT auto-confirm, ask the user first as it may cost credits
5. Thread reuse: Pass thread_id to continue a conversation — the Agent remembers prior context and can iterate on designs
6. Project required: All chat requests need a project_id — create one with project/save first or reuse an existing one
7. Mode affects cost: Fast mode (unlimited: false) uses credits for priority processing; unlimited mode (unlimited: true) is free but may queue

API Reference

• Documentation: https://lovart.info/lovart-api
• Developer Guide: https://lovart.info/lovart-ai-code