Skip to main content
POST
/
check-once
curl -X POST https://vibestream-production-64f3.up.railway.app/check-once \
  -H "Content-Type: application/json" \
  -d '{
    "youtube_url": "https://youtube.com/watch?v=abc123",
    "condition": "Are there more than 5 people visible?"
  }'

{
  "triggered": true,
  "explanation": "Approximately 8-10 people are visible in the frame.",
  "model": "gemini-2.0-flash",
  "frame_b64": null
}

​
Request

​
youtube_url
string
required
Live stream URL to check. Supports YouTube (youtube.com/watch?v=, youtu.be/) and Twitch (twitch.tv/channel) formats.
​
condition
string
required
Natural language condition to evaluate. Be specific for better accuracy.
​
model
string
default:"gemini-2.5-flash"
VLM model to use. Options: gemini-2.5-flash, gemini-2.0-flash, gpt-4o-mini, gpt-4o
​
include_frame
boolean
default:"false"
Include the captured frame as base64 in the response.
​
skip_validation
boolean
default:"false"
Skip livestream validation if already validated by /validate-url. Use this to avoid redundant validation calls when pre-validating URLs in the frontend.

​
Response

​
triggered
boolean
Whether the condition was met.
​
explanation
string
VLM explanation of what was observed in the frame.
​
model
string
The model that was used for analysis.
​
frame_b64
string
Base64-encoded frame (only if include_frame was true).
curl -X POST https://vibestream-production-64f3.up.railway.app/check-once \
  -H "Content-Type: application/json" \
  -d '{
    "youtube_url": "https://youtube.com/watch?v=abc123",
    "condition": "Are there more than 5 people visible?"
  }'

{
  "triggered": true,
  "explanation": "Approximately 8-10 people are visible in the frame.",
  "model": "gemini-2.0-flash",
  "frame_b64": null
}

​
Use Cases

Quick Checks: Use /check-once for one-off queries without starting a continuous job. 
# Is it currently raining?
curl -X POST https://vibestream-production-64f3.up.railway.app/check-once \
  -H "Content-Type: application/json" \
  -d '{"youtube_url": "https://youtube.com/watch?v=WEATHER_CAM", "condition": "Is it raining?"}'
Unlike /live-monitor, this endpoint is synchronous and blocks until the VLM returns a response. Typical response time is 2-5 seconds depending on the model.