sales-proposal-builder
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns or security vulnerabilities were detected across the skill's instructions or supporting scripts.
- [COMMAND_EXECUTION]: The skill executes a local Python script (
extract_brand_style.py) to extract branding tokens from user-provided files. This command execution is implemented securely by using platform-provided environment variables for path resolution (${CLAUDE_PLUGIN_ROOT}) and applying proper quoting to arguments to prevent shell injection. - [INDIRECT_PROMPT_INJECTION]: The skill includes functionality to ingest and process untrusted Office (PPTX/DOCX) and PDF documents. This attack surface is mitigated through a safety-hardened extraction script that enforces archive limits (e.g., maximum members and decompression ratios) and uses secure XML parsing. Additionally, the skill's operating principles instruct the agent to treat all external content as non-authoritative data.
- [SAFE]: The skill implements path sanitization for any file system operations involving user-provided strings (such as company names), effectively preventing directory traversal or path manipulation.
Audit Metadata