wechat-cell-writer

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFECREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill manages WeChat API credentials, including AppID and AppSecret, by retrieving them from a local configuration file at ~/.baoyu-skills/.env.
  • [COMMAND_EXECUTION]: A central workflow runner (run-workflow.js) orchestrates the execution of multiple local shell scripts and TypeScript files for article initialization, content validation, and final sanitization.
  • [REMOTE_CODE_EXECUTION]: The skill employs the Playwright library (screenshot-paper.ts) to automate a headless browser for navigating to external academic websites (e.g., PubMed, Nature) to capture document screenshots for article citations.
  • [EXTERNAL_DOWNLOADS]: As part of its setup and execution, the skill triggers the download of the Chromium browser binary via the Playwright framework.
  • [PROMPT_INJECTION]: The skill retrieves and processes untrusted data from external scientific sources and news media during its research phase.
  • Ingestion points: External content is aggregated into research.md during the retrieval step.
  • Boundary markers: Structured article templates and frontmatter are used to delimit generated content.
  • Capability inventory: The skill can perform shell execution, local Python processing, and browser automation.
  • Sanitization: The workflow includes mandatory sanitization (sanitize-article.sh) to strip URLs and compliance checks (validate-article.sh) to identify and flag prohibited terminology or exaggerated medical claims before publication.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 08:50 PM