The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external accounting and payment systems, creating a surface for indirect prompt injection.
Ingestion points: Revenue, cost, and transaction data are retrieved from connectors named ~~бухгалтерия (e.g., 1C, Moysklad) and ~~платежи (e.g., YooKassa, Tinkoff) as described in Step 1.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when processing the fetched business data.
Capability inventory: The skill is configured to use Bash, WebFetch, and Read tools, which provides a significant capability tier for an agent.
Sanitization: There are no instructions to sanitize or validate the content of the external data before it is processed or used in calculations.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform data analysis and potentially interact with local systems. While standard for this use case, it increases the impact of potential injection vulnerabilities.

price-check