Natural Language Video Search
PassAudited by ClawScan on May 10, 2026.
Overview
This skill’s behavior matches its video-search purpose, but it uses an external Gemini API key, uploads video chunks for embedding, stores a local vector index, and asks the user to install unreviewed code from GitHub.
Before installing, confirm you trust the GitHub repository and dependencies, use a limited Gemini API key, and only index video directories you intentionally want processed. Be especially careful with security-camera, dashcam, or Tesla footage because it may include faces, license plates, GPS, and location data.
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.
Installing the skill requires trusting code and dependencies that were not part of this artifact review.
The submitted skill is instruction-only and tells the user to fetch and install code from an external GitHub repository; the runnable code and dependency lock context are not included in the provided artifacts.
git clone https://github.com/ssrajadh/sentrysearch.git cd sentrysearch uv sync
Review the GitHub repository and dependency files before installing, and prefer pinned versions or a trusted release tag when possible.
A local project file may contain the user’s Gemini API key, which could be exposed if the directory is shared, backed up, or committed accidentally.
The skill requires a Gemini API key and documents that initialization stores it locally in a .env file, which is expected for this integration but is still credential handling.
This prompts for your key, writes it to `.env`, and validates it with a test embedding. You can also set `GEMINI_API_KEY` directly as an environment variable.
Use a dedicated Gemini key with appropriate limits, keep .env private, and avoid committing or sharing the installation directory.
Private dashcam, security, surveillance, or Tesla footage may be processed by an external provider and represented in a persistent local search index.
The skill sends video chunks to Gemini for embedding and persists searchable vectors locally. This is central to the purpose, but video footage and derived embeddings can be sensitive.
Video files are split into overlapping chunks... Each chunk is embedded as raw video using Gemini Embedding 2... Vectors are stored in a local ChromaDB database.
Index only footage you are comfortable sending to Gemini for embedding, avoid broad private folders unless intended, and understand where the local ChromaDB index is stored and how to delete it.