react-components

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly downloads and parses externally-provided Stitch HTML/assets (see SKILL.md workflow step 1, resources/stitch-api-reference.md describing signed downloadUrl, and scripts/fetch-stitch.sh which fetches arbitrary URLs) and uses that content to drive component extraction and generation, so untrusted third-party content can directly influence agent actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's runtime includes scripts/fetch-stitch.sh (invoked via npm script "fetch") which fetches an arbitrary Stitch downloadUrl (the parameter / signed downloadUrl referenced in resources/stitch-api-reference.md) and writes that remote HTML into the agent's workspace to be consumed as required input that directly influences the model's outputs—so external URLs fetched at runtime can control the agent's behavior.