The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several scripts, including deploy_scenescape.sh and bootstrap_deploy.py, to execute shell commands and orchestrate Docker Compose operations. These commands are responsible for pulling container images, managing persistent volumes, and starting the microservices required for the SceneScape stack.
[EXTERNAL_DOWNLOADS]: The skill downloads required assets, scripts, and OpenVINO detection models from official GitHub repositories and dedicated model delivery services. This includes sparse-cloning the SceneScape repository and fetching sample data from the author's official domains. All detected remote sources are associated with trusted organizations or well-known development services.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided RTSP stream URLs and camera identifiers as primary inputs for generating deployment configurations.
Ingestion points: Data enters the system via command-line arguments gathered in the Step 1 requirements phase and persisted in deploy-inputs.json.
Boundary markers: The skill incorporates validation logic within deploy_inputs.py and adapt_pipeline_config.py, which include URL parsing and identifier format checks.
Capability inventory: The skill has capabilities to write files to the deployment directory and execute shell commands via Docker and subprocesses.
Sanitization: RTSP URLs are validated using standard parsing libraries, and camera IDs are restricted to prevent character injection.
[PRIVILEGE_ESCALATION]: The skill manages file permissions using chmod to ensure that setup scripts are executable and that sensitive deployment secrets are protected with owner-only access (0600). These operations are routine for local deployment orchestration and do not attempt to acquire unauthorized system privileges.

scenescape-setup