braid-prompting
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (conversation history) to generate structured reasoning plans. This introduces a potential surface for indirect prompt injection where malicious input could influence the generated Mermaid diagram.
- Ingestion points: The
${conversationText}placeholder in the BRAID Generation Prompt section ofSKILL.md. - Boundary markers: The prompt template uses a 'Conversation:' label as a delimiter to separate instructions from the data.
- Capability inventory: The skill does not define or use any tools for network access, file system modification, or command execution; its output is restricted to Mermaid text.
- Sanitization: There are no instructions for sanitizing, escaping, or validating the conversation history before it is interpolated into the prompt template.
- [NO_CODE]: The skill consists solely of markdown-based instructions, definitions, and prompt templates. It does not contain any executable scripts, binary files, or tool configurations.
Audit Metadata