AI Bot System documentation
Real platform capabilities: supported devices, accepted event formats, delivery channels, and storage behavior.
Live endpoints and ports
The values below are generated from your current server configuration.
API ingest (JSON)
https://ai-bot-system.com/api/events
HTTP endpoint for cameras
https://ai-bot-system.com/camera
http://ai-bot-system.com:3001/
TCP endpoint for LPR
ai-bot-system.com:9000
Active ports: web 4000, https 4443, camera-http 3001, tcp 9000
Supported devices
What you can connect to the platform today.
Cameras and LPR devices
IP cameras, ANPR/LPR cameras, and camera integrations that send event payloads are supported.
Sensors and controllers
Any device that can push event data with identifiers (DeviceID/MAC/IP) can be connected.
Incoming formats
The ingest layer accepts JSON, multipart/form-data, x-www-form-urlencoded, text/xml/application/xml, and raw TCP camera payloads.
Data flow
Where incoming data goes and how it is processed.
Input endpoints
Events are received through HTTP ingest endpoints and optional TCP endpoint (port 9000 for LPR/camera integrations).
Delivery targets
Alerts are forwarded to Telegram bots and event metadata is written to dashboard event history.
Media handling
Snapshots can be attached to Telegram alerts and stored in local uploads or S3-compatible cloud storage.
Storage behavior
How the platform stores data and behaves near capacity.
Where data is stored
Events are stored in the database; media files are stored in local uploads or object storage depending on configuration.
When storage limit is reached
When user storage is full, new images are not saved to disk/cloud, but alert delivery and event logging continue.
Manual cleanup controls
You can disable media saving or clear accumulated alert media from dashboard profile settings.
How to configure cameras and AI
Minimal steps for stable event delivery and local model training.
Regular camera (non-LPR)
In camera HTTP action set Camera URL (HTTP), method POST, include DeviceID/MAC and a full frame. For multipart use snapshot/image/PictureData field names.
LPR / NVR camera
For LPR use TCP host:port endpoint. For NVR HTTP push use Camera URL (HTTP) and avoid sending duplicate MAC/CH/ALARM/TIME both as plain text and separate fields.
Local AI model training
Open Dashboard → Analytics: enable analytics per device, label events with True/False/Needs review in Recent analytics decisions, then click Start local model training.
Ready templates for NVR / AIBOX / and other HTTP Action capable devices
Copy a template and replace MAC/channel/keys with your values.
NVR / AIBOX: multipart + photo
Recommended option for alert events with an image (full frame).
POST https://ai-bot-system.com/camera Content-Type: multipart/form-data MAC=00:1C:B8:81:C7:06 CH=1 Event-Type=Intrusion Date-Time=2026/2/25 13:39:45 snapshot=@full_frame.jpg # Do not add a long duplicate text payload like: # "MAC: ...; CH: ...; ALARM: ...; TIME: ..." # These fields are already sent separately.
NVR / AIBOX: form-urlencoded
Fallback option without image if the camera cannot upload files.
POST https://ai-bot-system.com/camera Content-Type: application/x-www-form-urlencoded MAC=00:1C:B8:81:C7:06&CH=1&Event-Type=Intrusion&Date-Time=2026/2/25 13:39:45 # If a camera cannot upload files, # the event will still be logged, but for Telegram/analytics # it is recommended to send a full frame (snapshot/image).
Server/API JSON ingest
Template for server integration through signed ingest endpoint.
POST https://ai-bot-system.com/api/events
Headers:
X-API-Key: YOUR_DEVICE_API_KEY
X-Timestamp: UNIX_MS
X-Signature: HMAC_SHA256_HEX
Content-Type: application/json
{
"device_id": "NVR203_CH1",
"event_type": "intrusion",
"message": "Camera channel 1 intrusion",
"metadata": {
"MAC": "00:1C:B8:81:C7:06",
"CH": "1",
"Event-Type": "Intrusion",
"Date-Time": "2026/2/25 13:39:45"
}
}