The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script (price_scan.py) that manages the host's VPN connection. It has the capability to change the system's network exit point across different countries, which affects all network traffic on the host machine. These changes are performed based on marketplace requirements and user configuration.
[DATA_EXPOSURE]: The skill accesses sensitive local configuration files, specifically addresses.yaml and marketplaces.yaml, stored in the plugin data directory (e.g., ~/.local/share/claude-plugins/purchasing/). These files contain delivery address profiles and marketplace identifiers used to calculate landed costs and configure network routing.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to external marketplaces (such as Amazon and Newegg) and a public IP-geolocation endpoint to verify the VPN exit. While necessary for price retrieval and regional accuracy, this involves egressing data about the machine's public IP and intended shopping targets.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting untrusted data from external product pages and user-provided files like price-watchlist.yaml.
Ingestion points: External product HTML/data (via scraping) and the price-watchlist.yaml file in the repository root.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore embedded instructions within the scraped content.
Capability inventory: The agent has access to python3, Write, Edit, mkdir, and Bash commands, allowing for filesystem modification and script execution.
Sanitization: There is no mention of sanitization, filtering, or validation of the external content before it is processed or stored in markdown/JSON reports.

price-scan