nsfc-writer

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety guidelines, or extract system prompts were found. The instructions are strictly task-oriented and professional.
  • [DATA_EXFILTRATION]: The skill does not access sensitive local file paths (such as .ssh or .aws/config) or hardcode any credentials. It utilizes literature search tools (PubMed, Academic Search) to fetch public research data, which is consistent with its stated purpose.
  • [REMOTE_CODE_EXECUTION]: The skill mentions using MCP (Model Context Protocol) tools for PubMed and Academic Search, which are well-known and trusted services. There are no instructions to download and execute external scripts or install unverified packages.
  • [OBFUSCATION]: The content is clear and uses standard Markdown formatting. No hidden content, Base64 encoding of commands, zero-width characters, or homoglyphs were detected.
  • [COMMAND_EXECUTION]: There are no instances of subprocess spawning, shell command execution, or dangerous system-level operations.
  • [INDIRECT_PROMPT_INJECTION]: While the skill ingests untrusted user data (research drafts and topics), it does so within the context of a writing assistant. It lacks the high-privilege capabilities (like file system writes or unauthorized network requests) that would make this surface critical.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 01:10 PM