gemini-count-in-video
ReviewAudited by ClawScan on May 10, 2026.
Overview
The visible artifacts are coherent for a Gemini video-counting skill, but users should notice that it needs a Gemini API key and uploads selected or batched video footage to Google.
Before using this skill, confirm you are allowed to share the videos with Google Gemini, set GEMINI_API_KEY securely, use a narrowly scoped folder for batch jobs, and monitor API cost and data-retention implications.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Users may need to provide an API key that can incur charges or access their Google AI account quota.
The skill expects a Gemini API key. This is normal for a Google Gemini integration, but it grants access to a user's billed API account and is not declared in the provided registry requirements.
client = genai.Client(api_key=os.getenv("GEMINI_API_KEY"))Use a dedicated Gemini API key with appropriate restrictions, store it securely, and monitor API usage and costs.
Video footage may contain personal or sensitive information and will be processed by Google rather than staying entirely local.
The documented workflow uploads video files, including surveillance footage, to the Gemini File API for analysis. This is central to the skill's purpose but creates an external data-sharing boundary.
Use the File API upload flow for larger videos (most surveillance footage).
Only analyze videos you are authorized to share, review Google Gemini data handling and retention terms, and avoid uploading unnecessary sensitive footage.
If the input folder is not curated, more videos than intended could be sent to Gemini and billed for processing.
The batch example processes every matching video in a directory. This is purpose-aligned for batch counting, but users should understand that all matching files in the chosen folder may be uploaded and analyzed.
for filename in os.listdir(video_dir): ... myfile = upload_and_wait(client, video_path)
Use a dedicated input folder, review the file list before running batch processing, and avoid pointing the workflow at broad or private directories.