writing-revision

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the draft text it processes.
  • Ingestion points: The skill ingests user-supplied draft text as its primary input for the three-pass revision workflow defined in SKILL.md.
  • Boundary markers: There are no explicit instructions or requirements to use delimiters (such as triple backticks or XML tags) to separate the untrusted draft content from the revision instructions.
  • Capability inventory: The skill involves local file system operations, specifically writing findings to analysis files (e.g., writer-pass1-clutter-analysis.md) and reading them back to perform improvements, which could be leveraged if an injection is successful.
  • Sanitization: The instructions do not include steps to sanitize the draft text or warn the agent to ignore any embedded commands or formatting instructions within the user's content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:07 PM
Security Audit — agent-trust-hub — writing-revision