The Agent Skills Directory

[NO_CODE]: The skill package is composed entirely of Markdown files providing instructions, stylistic references, and before/after examples. There are no executable scripts, binaries, or configuration files that trigger automated code execution.
[SAFE]: The skill performs only text-to-text transformations based on defined stylistic rules. It does not request access to tools, the network, or the local file system beyond reading its own reference files for instruction context.
[INDIRECT_PROMPT_INJECTION]: The skill functions by ingesting and editing user-provided prose. While it lacks explicit delimiters to isolate user content within its instructions, the complete absence of exploitable capabilities—such as shell access, file writing, or network requests—ensures that any malicious instructions embedded in the processed text cannot result in system compromise.
[METADATA_POISONING]: The author name specified in the skill's metadata and README ('Hardik Pandya') differs from the provided vendor context ('persiyanov'). While technically a discrepancy, the content is consistent with a legitimate writing style guide and presents no evidence of deceptive intent or hidden functionality.

stop-slop