x-ai-digest
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION] (LOW): Indirect Prompt Injection Surface. The skill ingests untrusted content from X platform posts and uses an LLM to generate summaries and 'reply suggestions'.
- Ingestion points:
references/scraping-guide.mdextracts tweet content using the[data-testid="tweetText"]selector. - Boundary markers: No specific delimiters or safety instructions are used in
assets/digest-template.mdto prevent the LLM from obeying instructions embedded within scraped tweets. - Capability inventory: The skill performs file system writes (Markdown, JSON, PNG) and network operations (via browser automation).
- Sanitization: No sanitization or filtering of scraped tweet text is implemented before interpolation into the prompt for digest generation.
- [EXTERNAL_DOWNLOADS] (LOW): Dependency on external tools and binaries.
- The skill requires the
dev-browserskill andplaywrightpackage. - It executes
npm run start-extensionand launches a Chromium instance to render and screenshot HTML templates. - While Playwright is a standard library, it involves the execution of external browser binaries.
Audit Metadata