The Agent Skills Directory

[PROMPT_INJECTION]: The skill does not contain any instructions attempting to bypass safety guardrails, override previous instructions, or extract system prompts. The directive language is focused on scientific writing workflows.
[DATA_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or network operations were identified. The use of placeholders like [DOI] or [ACCESSION] follows standard documentation practices for templates.
[OBFUSCATION]: Analysis of the skill body and reference files, including the Chinese-language alignment guide, found no evidence of Base64 encoding, zero-width characters, or homoglyph-based URL spoofing. The non-Latin text is legitimate instructional content for bilingual authors.
[REMOTE_CODE_EXECUTION]: The skill does not include any commands for downloading external scripts, installing third-party packages, or executing remote code.
[COMMAND_EXECUTION]: There are no shell command patterns or dynamic execution markers found in the documentation or reference files.
[INDIRECT_PROMPT_INJECTION]: The skill defines a clear ingestion surface for manuscript drafts and repository information. However, the skill lacks capabilities (such as writing to the file system or making network requests) that would make it vulnerable to exploitation through data ingestion. No evidence of self-referential safety claims was found.

data-availability