The Agent Skills Directory

[SAFE]: The skill performs legitimate analysis of subscription data using the transaction_search and anomaly_detect tools. It aggregates transaction patterns to provide cost visibility and cancellation recommendations to the user without external data transmission.
[INDIRECT_PROMPT_INJECTION]: The skill processes transaction descriptions (merchant names) which originate from external sources. While this presents a surface for indirect injection, the skill lacks the capabilities required for exploitation.
Ingestion points: Merchant descriptions and transaction records ingested via the transaction_search tool in SKILL.md.
Boundary markers: None explicitly defined in the workflow instructions.
Capability inventory: The skill is restricted to data aggregation and presentation; it does not utilize network operations, file-write capabilities, or shell execution.
Sanitization: No explicit sanitization of merchant strings is performed, but the results are only used for cost calculation and table presentation.

subscription-audit