The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python script scripts/consistency_audit.py designed for local execution to perform document auditing. The script utilizes standard file I/O and text processing techniques. It does not spawn shells or execute arbitrary external commands based on untrusted input.
[EXTERNAL_DOWNLOADS]: The references.md file contains several informational links to established development tools, official documentation (e.g., Mermaid, Style Dictionary, W3C Design Tokens), and public GitHub repositories for research purposes. These references are documented neutrally and do not involve automated or unsafe background downloads.
[DATA_EXFILTRATION]: The auditing script reads local project materials provided as command-line arguments. It does not contain any networking capabilities (e.g., requests, urllib) or patterns suggesting the transmission of data to remote servers.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text files to check for consistency. While the skill ingests untrusted data from these files, its capabilities are restricted to generating local reports and providing text-based recommendations. There is no evidence that malicious content within audited materials can trigger unintended system actions or bypass safety filters.

light-consistency