- Home
- Skills
- Testing & Debugging
- Cry Wolf: Stop Frigate False Alarms
More screenshots
Works with the AI Agents you already use
Cry Wolf: Stop Frigate False Alarms
Stop a chair, a plant or the street outside from setting off your Frigate alerts. Your agent groups past detections by box, shows what caused them, and estimates what a filter would drop before you change anything.
$7
Cry Wolf: Stop Frigate False Alarms
Example session with this skill installed
Frigate keeps alerting 'person' on my back yard camera at night and I'm about to draw a motion mask over that corner. I clustered the last 300 person events by box. The top row: count 61, box 0.3,0.7,0.3,0.3, pixel ratio 1.07, area 7.9%, scores 65-91%, 61 detections in 47 minutes, box jitter 0.004. Should I draw the motion mask? What would you do?
- Read your context and instructions
- Compiled the cry wolf
Summary
No, skip the motion mask. It only stops motion in that corner from starting detection, so the spot would still fire. It can also hurt tracking of real people there.
Your numbers point to one static object. Nobody holds a box to 0.004 jitter across 61 detections in 47 minutes. It's probably furniture, a plant or similar. Raising threshold won't help either, since it peaks at 91% and the change would hit real people too.
What the answer covers
- Confirm what the object is from snapshots.
- Apply a targeted filter for that one spot.
- Check your other 239 events first, so real people aren't dropped.
- Confirm after a week that alerts stopped and real people still show up.
The skill works out the filter for your numbers with the bundled tool.
Connects securely to your tools. The creator never sees your data.
About this skill
Two cameras with the same complaint often need opposite fixes. A chair needs an object filter mask. A busy pavement needs a required zone. Guess wrong and you either keep the alerts or lose real people.
Your agent will
- Pull your camera's recent events and group them by bounding box
- Mark likely static objects using how often and how steadily a box repeats
- Show whether small boxes high in the frame point at the street
- Replay a candidate filter against history and list the events most likely to be real people
- Warn you before a filter removes too much
How it was tested
We ran the script on 1,655 real events across five cameras on Frigate 0.17. Testing on real data changed the script: it now separates a parked object from people walking the same path, and it won't report a safety number that the filter itself guarantees.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Replay uses one stored box per event while Frigate filters every frame, so treat it as an estimate.
History holds only what passed your existing filters.
Open a snapshot before you trust a STATIC? flag. A parked car looks the same.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 8 days ago
- Software
- Frigate with API access
Python 3
Permissions
Allowed Hosts
File Scopes
API calls go only to the Frigate address the user gives. The script reads one local JSON file and makes no network calls.