The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a third-party repository (github.com/challengekim/iros-registry-automation) to obtain automation scripts. While the source is external, the skill mitigates risk by specifying a pinned commit SHA (7c6924b2ff88d693a12556659188cb91041e5097) to ensure integrity and prevent the execution of unreviewed updates.
[COMMAND_EXECUTION]: The workflow relies on the execution of multiple Python scripts (e.g., iros_cart.py, iros_download.py, iros_wizard.py) and command-line tools like playwright. These scripts interact with the local file system and automate browser sessions to perform batch tasks on the registry website.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from registration lists and registry-issued PDF documents to generate reports and map business information. This ingestion of untrusted data represents an attack surface for indirect prompt injection.
Ingestion points: Data is read from user-provided files such as corp-input.json, realty_list, and content extracted from downloaded registration PDFs.
Boundary markers: The instructions recommend using a restricted temporary directory (chmod 700) for data processing, though it lacks explicit delimiters or instructions for the agent to ignore potentially malicious embedded text in the PDFs.
Capability inventory: The skill involves executing shell commands, writing files to the local disk, and using the Playwright browser automation framework.
Sanitization: No specific sanitization or validation logic is described for the text content parsed from external PDF registry documents before it is used in subsequent agent responses or reporting tasks.

legal-iros-registry-automation