The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill provides instructions involving administrative commands that modify system-wide settings.
Instructions in references/troubleshooting.md recommend sudo ufw disable to resolve WebRTC connectivity issues, which significantly reduces the system's security posture.
The references/service-setup.md file includes commands like sudo apt install systemd-timesyncd and sudo systemctl restart systemd-timesyncd for managing system time services.
[DYNAMIC_EXECUTION]: The skill supports the udfloader GStreamer element, as documented in SKILL.md and references/api-and-pipelines.md.
This element is designed to load and execute custom Python User Defined Functions (UDFs) within the video pipeline, representing a runtime code execution capability.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, potentially untrusted sources.
Ingestion points: The skill accepts video files and network streams via source.uri in the REST API request body (e.g., file:///path/to/video.avi).
Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the content being processed from the agent's instructions.
Capability inventory: The skill employs docker compose to manage containers and curl for interacting with the microservice API.
Sanitization: There is no mention of sanitization or validation of the media content before it is processed by the AI-integrated pipeline.
[EXTERNAL_DOWNLOADS]: The skill references external code and container images from established sources.
The skill fetches configuration and code from the open-edge-platform GitHub repository (git clone https://github.com/open-edge-platform/edge-ai-libraries.git).
The service uses official container images from Intel, a well-known technology vendor (docker pull intel/dlstreamer-pipeline-server:latest).

dlsps-user