The Agent Skills Directory

[SAFE]: The skill functions as a logic-based calculator that processes user-provided inputs against internal reference documents (references/coupon-rules-jd.md, references/coupon-rules-pdd.md). It does not utilize any tools for network communication, local file writing, or administrative command execution.
[SAFE]: A dedicated policy file (references/no-fabrication.md) establishes a robust integrity framework, explicitly prohibiting the generation of unverified pricing, platform links, or promotional data. This significantly reduces the risk of the agent providing deceptive or hallucinated information.
[SAFE]: Indirect Prompt Injection risk is minimized through clear operational boundaries. Ingestion points for untrusted user data (coupon names and prices) are managed by strict instructions to verify data via screenshots or full official names. The skill possesses no dangerous capabilities such as eval(), exec(), or subprocess spawning that could be triggered by malicious inputs.

coupon-stacker