Any plans to integrate any sort of notofcations (say email / webhooks) from Frigate natively ? #20879
-
|
The frigate AI stuff is coming along nicely. I have frigate adding the face and the AI description to its alerting, but to get that added to email / webhooks, I also have to have home assistant doing the same thing - using up the AI credits. I was wondering if you plan to add an option so that the AI description can be sent out (ideal email / webhooks to discord as that is what I use in HA). |
Beta Was this translation helpful? Give feedback.
Replies: 7 comments 5 replies
-
|
Frigate supports native notifications using the WebPush protocol, which can deliver notifications directly to web browsers, but does not natively support sending notifications via email or webhooks out-of-the-box(1). The configuration allows you to set an email address, but this is only used for push service registration, not for sending standard email alerts(1), (2). For advanced workflows—such as integrating the AI description, face data, or other alert details into notifications sent to email, Discord, or webhooks—you currently need to use Home Assistant or another automation platform. Frigate provides MQTT topics (such as frigate/tracked_object_update and frigate/events) that can be used to trigger automations in Home Assistant, which can then forward notifications containing the AI description to your preferred channels(3), (4), (5), (6). This does mean that if Home Assistant is also calling external AI services, you may use additional AI credits. There is no documented plan or feature for Frigate to natively send notifications with AI descriptions via email or webhooks at this time. All current examples and guides rely on external automation (like Home Assistant) to relay and enrich Frigate alerts to email, Discord, or similar services(1), (3), (4), (6). 📚 Sources:
Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
I read some of the doco that the AI reference and noticed this below - which could halve my AI requests if I can pull this out of mqtt in HA and email / webhook that instead. Have to have a conversation with your AI on how to do that as I get completely lost with the mqtt stuff. Frigate provides an MQTT topic, frigate/tracked_object_update, that is updated with a JSON payload containing event_id and description when your AI provider returns a description for a tracked object. This description could be used directly in notifications, such as sending alerts to your phone or making audio announcements. If additional details from the tracked object are needed, you can query the HTTP API using the event_id, eg: http://frigate_ip:5000/api/events/<event_id>. |
Beta Was this translation helpful? Give feedback.
-
|
So - I connected via cloudflare so using ssl to turn on notifications as per doco - and seemed to have stuffed the frigate system now - I just get a spinning circle on startup. Cant even get logs up trhough the gui - but below is what was shown via the docker container logs. I am able to go into the config of the docker container and removed all references to notifications, saved and restarted the container. All back working again now. 025-11-11 10:58:13.337182395 2025/11/11 10:58:13 [error] 210#210: *216 connect() failed (111: Connection refused) while connecting to upstream, client: 192.168.1.10, server: , request: "GET /ws HTTP/1.1", subrequest: "/auth", upstream: "http://127.0.0.1:5001/auth", host: "10.10.10.10:5000" |
Beta Was this translation helpful? Give feedback.
-
|
write a home assistant automation so that is a person is detected in binary_sensor.pathway_person_occupancy, then wait for a person to be detected in binary_sensor.doorway_person_occupancy, and then send a discord message via rest_command.discord_message_frigate with the frigate person name if known, the frigate face confidence percentage and the frigate generated ASI description. |
Beta Was this translation helpful? Give feedback.
-
|
To accomplish this automation with Home Assistant and Frigate, you will need to:
The sources show that the frigate/tracked_object_update MQTT topic provides a payload like: {
"type": "description",
"id": "event_id",
"description": "The car is a red sedan moving away from the camera."
}You can then use the event_id to query the HTTP API for more details: The automation logic would be as follows (pseudocode, as exact YAML depends on your Home Assistant setup):
You can find further details about the MQTT topic structure and API usage at:
📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
There is no directly quoted, complete Home Assistant automation YAML in the provided sources that matches your scenario. However, the sources do describe the required steps and logic, including how to use MQTT triggers, REST API calls, and Home Assistant sensors to achieve your goal with Frigate and GenAI descriptions(1)(2)(3)(4). Check out these potentially useful sources: Would you like a general answer with an example automation based on my knowledge instead? 📚 Sources:
Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.
-
|
There is no complete, directly quoted Home Assistant automation YAML covering your scenario in the provided sources, so I cannot fully verify or correct every line as a supported answer. However, your core logic aligns with how Frigate publishes face and GenAI description updates to the frigate/tracked_object_update MQTT topic. The payload structure you’re checking for (type: "face" or "description", camera, name, score, description) matches the documented Frigate MQTT messages for face recognition and generative AI description updates(1)(2). Waiting for occupancy sensors and then dispatching a Discord message via rest_command is a standard Home Assistant approach. For MQTT triggers, ensure that your topic and payload matching logic are correct. The camera value in the MQTT payload must match exactly what Frigate publishes (e.g., "front_door_cam" vs "Front-Door")—check your Frigate camera names for exact case and spelling(1). The value_template and payload usage in your wait_for_trigger should reflect the actual MQTT payload structure. The variables person_name, face_score, and asi_description are set according to the presence of those fields in the MQTT message payload, which is consistent with the documented examples(1)(2). Your approach to send a message with these variables via rest_command.discord_message_frigate is also appropriate. Summary: Your logic is sound and matches how Frigate’s MQTT topics for face and generative AI description updates work. Just ensure your camera names and MQTT topic matching are exact, and that the payload fields in your automation correspond to the actual Frigate MQTT messages(1)(2). 📚 Sources: Have another question? Just tag @inkeep. |
Beta Was this translation helpful? Give feedback.

Just to close this off, I worked for a while with some AI's to get this data, and we ended upon three separate automations -
One runs as a passive listener on AI DEescriptions on the front door camera and updates a sensor.
One runs on the last face detection only and updates a sensor. This send discord message if face is recognised or not.
One runs on the last AI description and updates a sensor. This send discord message but only if one is generated.
These are also added to a markdown card with the alert times for the dashboard.
The automations are shown below if this helps anyone else - The AI tried its hardest to do the face and description in one automation but it was always one step …