The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a bulk import feature (POST /v3/session/imports/) that fetches verification data from a user-specified remote URL (source_file_url). This creates a surface for indirect prompt injection, where an attacker could provide a URL pointing to a malicious file containing instructions designed to override the agent's behavior. 1. Ingestion points: The source_file_url parameter and the resulting data fetched from the provided remote host. 2. Boundary markers (absent): There are no instructions or delimiters defined to ensure the agent ignores natural language instructions embedded within the imported data. 3. Capability inventory: The skill has extensive administrative capabilities, including the ability to approve/decline verification sessions, modify blocklists, and configure webhook destinations. 4. Sanitization (absent): The provided documentation and scripts do not describe or implement validation or sanitization of the imported file content.
[EXTERNAL_DOWNLOADS]: The skill documentation and accompanying Python scripts facilitate network communication with various *.didit.me subdomains for API functions and checkout.stripe.com for payment processing. These are legitimate endpoints belonging to the skill vendor or well-known services and are documented neutrally.
[COMMAND_EXECUTION]: The skill's utility scripts (setup_account.py, etc.) require users to pass sensitive information like passwords and API keys as command-line arguments. While standard for many CLI tools, this practice can expose credentials to other users or processes on the same system monitoring the process list.

didit-verification-management