メニュー
Home Assistant camera integration and Frigate NVR patterns.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | camera-nvr |
| description | Home Assistant camera integration and Frigate NVR patterns. |
Home Assistant camera integration and Frigate NVR patterns.
# frigate.yaml
mqtt:
enabled: true
host: mosquitto
user: frigate
password: "{FRIGATE_MQTT_PASSWORD}"
detectors:
coral:
type: edgetpu
device: usb
cameras:
front_door:
ffmpeg:
inputs:
- path: rtsp://user:pass@192.168.1.100:554/stream1
roles:
- detect
- record
- path: rtsp://user:pass@192.168.1.100:554/stream2
roles:
- rtmp
detect:
width: 1920
height: 1080
fps: 5
objects:
track:
- person
- car
- dog
filters:
person:
min_area: 5000
threshold: 0.7
zones:
porch:
coordinates: 0,1080,500,1080,500,0,0,0
objects:
- person
record:
enabled: true
retain:
days: 7
mode: motion
events:
retain:
default: 30
go2rtc:
streams:
front_door:
- rtsp://user:pass@192.168.1.100:554/stream1
# Generic camera
camera:
- platform: generic
name: Front Yard
still_image_url: http://192.168.1.100/snapshot.jpg
stream_source: rtsp://user:pass@192.168.1.100:554/stream
- platform: onvif
host: 192.168.1.103
port: 80
username: admin
password: !secret camera_password
automation:
- alias: "Camera - Person Alert"
trigger:
- platform: state
entity_id: binary_sensor.front_door_person
to: "on"
condition:
- condition: state
entity_id: input_boolean.camera_alerts
state: "on"
action:
- service: notify.mobile_app
data:
title: "Person Detected"
message: "Person at front door"
data:
image: /api/frigate/notifications/front_door/snapshot.jpg
tag: camera_alert
actions:
- action: "VIEW_CAMERA"
title: "View"
# Frigate recording settings
record:
enabled: true
retain:
days: 7
mode: motion
events:
pre_capture: 5
post_capture: 5
retain:
default: 30
mode: active_objects
# Picture glance card
type: picture-glance
title: Front Door
camera_image: camera.front_door
camera_view: live
entities:
- binary_sensor.front_door_person
- binary_sensor.front_door_motion
# Frigate card
type: custom:frigate-card
cameras:
- camera_entity: camera.front_door
frigate:
url: http://frigate:5000
client_id: front_door
menu:
buttons:
frigate: true
fullscreen: true
download: true
live:
provider: webrtc-card
# PTZ service calls
service: onvif.ptz
data:
entity_id: camera.ptz_camera
move_mode: RelativeMove
pan: 0.1 # -1 to 1
tilt: 0.1
zoom: 0.1
# Preset
service: onvif.ptz
data:
entity_id: camera.ptz_camera
preset: "1"
| Use Case | Camera Type | Resolution |
|---|---|---|
| Doorbell | Video doorbell | 1080p+ |
| Driveway | Bullet camera | 4K |
| Backyard | PTZ camera | 1080p |
| Indoor | Cube/dome | 1080p |
# Coral USB
detectors:
coral:
type: edgetpu
device: usb
# Coral M.2/PCIe
detectors:
coral:
type: edgetpu
device: pci
# Intel OpenVINO
detectors:
openvino:
type: openvino
device: GPU
| Brand | URL Format |
|---|---|
| Hikvision | rtsp://user:pass@ip:554/Streaming/Channels/101 |
| Dahua | rtsp://user:pass@ip:554/cam/realmonitor?channel=1 |
| Reolink | rtsp://user:pass@ip:554/h264Preview_01_main |
| Amcrest | rtsp://user:pass@ip:554/cam/realmonitor?channel=1 |
| Ubiquiti | rtsp://ip:7447/camera_id |
Select and coordinate multi-agent teams (topology kits, role-based squads, lifecycle, worktree isolation). Use this skill whenever launching parallel agents, designing a review board, running a debug council, scheduling an orchestrator-workers team, configuring agent tool restrictions, or deciding between solo and team execution. Triggers on: "launch a team", "parallel agents", "review board", "debug council", "architect-implementer-reviewer", "swarm", "multi-agent", "subagents for X", "team topology", "agent lifecycle".
Select and wire an agentic design pattern (reflection, prompt chaining, routing, parallelization, orchestrator-workers, evaluator-optimizer, ReAct, blackboard) into the 5-layer Claude Code stack. Use this skill whenever deciding how to structure a multi-step task, whether to spawn subagents, how to run parallel review, or when to use which pattern. Triggers on: "which pattern", "orchestrate", "parallel review", "self-review", "chain of thought", "eval-optimizer loop", "blackboard", "ReAct", "how to decompose this task".
Auto mode permission handling — classifier-based approvals, PermissionDenied hook, defer permissionDecision, and autonomy profiles for hands-off Claude Code usage
Configure Claude Code's autonomous operating mode — profile selection (conservative, balanced, aggressive, unattended-review), permission blocks, and the three gate agents (planner, verifier, reviewer). Use this skill whenever enabling autonomous mode, switching profiles, tightening permissions for production branches, or setting up unattended execution. Triggers on: "autonomy", "unattended mode", "auto-approve", "permission mode", "autonomy profile", "gates", "/cc-autonomy", "planner verifier reviewer", "let claude run on its own".
Manage Claude Code's context window — token arithmetic, /compact strategy, anchor preservation, progressive loading, session analytics. Use this skill whenever a session gets long, context approaches limits, after /compact, when deciding what to load into CLAUDE.md vs leave in references, or when analyzing session cost/token usage. Triggers on: "context full", "compact", "too many tokens", "budget", "session analytics", "save tokens", "context window", "/compact strategy".
Evidence-driven deep analysis for hard coding problems — architecture decisions, root-cause investigation, high-stakes refactor planning, performance bottleneck isolation. Use this skill whenever the user asks for "the best approach", a "deep analysis", "root cause", "principal engineer review", or runs /cc-intel. Also triggers on hard debugging questions, major architectural choices, tricky performance problems, or any task where a hypothesis tree and evidence table matter more than a fast answer.