markdown-fixer
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references authoritative technical specifications and guidelines from established organizations, including the W3C (WCAG 2.2), CommonMark, and Unicode. These references are used for instructional context and do not involve executable code downloads.
- [COMMAND_EXECUTION]: The agent is designed to modify markdown files on the local filesystem to improve accessibility (e.g., fixing heading levels, tables, and link descriptions). This file-write capability is restricted to a single-pass edit after a review gate, aligning with its primary purpose.
- [PROMPT_INJECTION]: As the skill processes untrusted markdown content that may contain instructions, it has an inherent surface for indirect prompt injection. However, the instructions provide a highly structured logic for text transformations (regex-like patterns) and mandate a 'Phase 3 review gate' and human confirmation, which significantly mitigates the risk of the agent following malicious instructions embedded in processed documents.
Audit Metadata