infographics

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill documentation exposes an insecure pattern by offering a --api-key CLI option (which encourages embedding an API key verbatim in commands) even though it also documents the safer environment-variable approach; this means an LLM prompted to produce command examples could be asked to output secrets directly.

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

• Third-party content exposure detected (high risk: 0.75). When --research is enabled, the script calls Perplexity Sonar Pro and then injects the returned researched text (research_data['content']) into the generation prompt that is sent to the LLM, so outsider-authored free text from a third-party research service can enter the LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill makes runtime API calls to OpenRouter (used as https://openrouter.ai/api/v1) whose model responses (Perplexity Sonar research output and Gemini 3 Pro review critiques) are injected back into the generation prompts and drive iterative regeneration, so external content from that URL directly controls prompts and is a required runtime dependency.