marketing
Fail
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONDYNAMIC_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Automated scanners have flagged a repository file, 'channels/community/directory-submission/references/submission-tracker-template.csv', as malicious (FileRepMalware). While primarily a data tracking template, its presence triggers a critical detection alert.
- [EXTERNAL_DOWNLOADS]: The file 'channels/community/directory-submission/references/directory-list.md' includes a reference to 'https://fitprofessionals.net', which has been flagged as a malicious URL by automated scanners. The guide encourages users to submit product information to this and other directories.
- [PROMPT_INJECTION]: Several guides include strict behavioral constraints designed to override the agent's instructions. For example, 'paid-ads/platforms/google-ads/references/rsa-output-spec.md' uses mandatory language like 'output MUST comply' and 'enforce before responding'. Similarly, 'paid-ads/platforms/google-ads/references/audit-guardrails.md' defines 'Hard stops' that instruct the agent to 'Refuse' specific user requests and treat these rules as 'response contracts, not suggestions'.
- [INDIRECT_PROMPT_INJECTION]: The skill frequently ingests untrusted data from the web, including website listings, user reviews, and social media posts (documented in 'aso/GUIDE.md' and 'prospecting/references/demand-signals.md'). Although the skill includes instructions to treat this as data and ignore embedded commands, the systematic ingestion of arbitrary external HTML and text content constitutes a significant vulnerability surface for indirect prompt injection.
- [DATA_EXFILTRATION]: The skill recommends using 'curl' to fetch data from domains outside the whitelist, such as 'reddit.com', 'news.ycombinator.com', and 'public.api.bsky.app'. While used for social listening, these constitute network operations to third-party services.
- [DYNAMIC_EXECUTION]: The workflows for motion video generation ('paid-ads/ad-creative/references/motion-video-ads.md') and product demo recording ('content/video/references/product-demo-recording.md') rely on the dynamic execution of generated code and templates through command-line tools like Hyperframes, Remotion, and Playwright.
Recommendations
- CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
- AI detected serious security threats
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata