aws-wechat-article-review
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the execution of local scripts to automate part of the review process.\n
- Evidence: It calls
{python} {baseDir}/../aws-wechat-article-writing/scripts/write.py checkto perform machine-led self-checks on article drafts.\n - Evidence: It invokes
{python} {baseDir}/../aws-wechat-article-writing/scripts/write.py strip-citationsto sanitize articles by removing internal source-tracking metadata before finalization.\n- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted user data, creating a potential surface for indirect prompt injection.\n - Ingestion points: The agent reads article content from
draft.md,article.html, and metadata fromarticle.yaml.\n - Boundary markers: The instructions do not mandate the use of delimiters (e.g., XML tags or unique separators) to isolate user content from the agent's logic.\n
- Capability inventory: The skill is restricted to local file read/write operations and the execution of suite-specific Python scripts.\n
- Sanitization: The skill incorporates a specific sanitization step to strip citation strings from the text via a regular-expression-based script.
Audit Metadata