grey-haven-creative-writing

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Prompt Injection] (SAFE): The prompt templates contained within the skill (e.g., for believability audits and prose style analysis) are strictly task-oriented. There are no instructions designed to bypass safety filters or override the system prompt. Instructions like "AUDIT ONLY -- NO EDITS" act as functional constraints for the agent's logic.
  • [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive file paths, or network exfiltration patterns (such as curl or fetch to unknown domains) were found. The skill uses variable placeholders like {{MANUSCRIPT_PATH}} for local file operations, which is appropriate for its intended use case.
  • [Remote Code Execution] (SAFE): The skill contains no executable scripts or commands that download and run remote code. The workflows are entirely focused on text analysis and transformation within the agent's operational environment.
  • [Indirect Prompt Injection] (SAFE): While the skill is designed to ingest and analyze untrusted user data (manuscripts), its capabilities are restricted to generating reports, scores, and structured text. It lacks high-risk capabilities like network access or shell execution that would be necessary for an indirect injection attack to be impactful.
  • [Obfuscation] (SAFE): No base64 encoding, zero-width characters, or homoglyphs were detected in the analyzed files. All instructions and guides are in clear, human-readable Markdown.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:09 PM