The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses instructional language to guide the agent in supply chain analysis (e.g., 'IRON LAW: End-to-End'). These are business logic constraints rather than attempts to bypass AI safety filters or override system instructions.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive API keys, or private file paths were detected. The skill uses placeholders like '{Company}' and '{gap}' for its analysis output.
[OBFUSCATION]: No Base64, hex-encoded strings, or hidden Unicode characters intended for malicious use were found. The character rendering issue in the example section (貓) appears to be a standard encoding error for a local brand name ('黑貓') rather than an obfuscation attempt.
[EXTERNAL_DOWNLOADS]: The skill does not perform any network operations, external downloads, or package installations. All references are local files within the skill folder.
[COMMAND_EXECUTION]: The skill is composed entirely of Markdown documentation. There are no scripts, subprocess calls, or shell commands provided for execution.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process user-provided business scenarios, it does not utilize any high-risk tools or capabilities (like file writes or network access) that would make it vulnerable to indirect injection attacks. The risk surface is limited to text-based analysis.

biz-supply-chain