wonda-cli

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt contains multiple examples that pass API tokens, cookies, and plaintext passwords directly on the command line or via heredoc/flags (e.g., --auth-token, --ct0, --li-at-value, --jsessionid-value, --password-stdin <<< "hunter2"), which would require an LLM to echo secret values verbatim into commands or code — an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and processes public, user-generated content (e.g., wonda stealth-browser record and wonda brand extract for arbitrary URLs, wonda scrape social / wonda x search / wonda reddit search / wonda linkedin search and scrape video) and the workflow instructions require the agent to read and act on those external pages/posts (e.g., "scrape viral reel → generate 5 UGC", "reddit-subreddit-intel", brand extract from a URL), so untrusted third-party content can directly influence tool decisions and actions.