markstream-custom-components
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The instructions are strictly limited to technical implementation details for frontend components. There are no attempts to bypass safety filters, override system instructions, or extract sensitive prompts.
- [DATA_EXFILTRATION]: No network operations, file access commands, or credential harvesting patterns were found. The skill does not interact with sensitive local files or external endpoints.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any external package installations or download-and-execute operations. It provides guidance on using local imports from the expected 'markstream' package ecosystem.
- [OBFUSCATION]: All content is in plain text. No Base64, zero-width characters, homoglyphs, or other encoding techniques were used to hide malicious payloads.
- [INDIRECT_PROMPT_INJECTION]: Although the skill focuses on rendering Markdown and custom tags (which are common attack surfaces), it explicitly recommends using 'trusted custom tags' and 'sanitized HTML props'. It provides a structured architectural approach to rendering without introducing dangerous interpolation or bypasses.
- [DYNAMIC_CONTEXT_INJECTION]: The skill does not utilize any pre-execution shell commands or dynamic context placeholders.
Audit Metadata