stitch-react-components

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

• Third-party content exposure detected (high risk: 0.85). The required workflow fetches Stitch screen HTML from htmlCode.downloadUrl (a runtime-provided URL) via bash scripts/fetch-stitch.sh "<htmlCode.downloadUrl>" ..., and that downloaded HTML is then parsed/ingested into the agent’s LLM context—this is outsider-authored free text originating from Stitch’s content.

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

• Potentially malicious external URL detected (high risk: 0.90). This skill calls Stitch MCP to obtain an htmlCode.downloadUrl (often a Google Cloud Storage URL such as https://storage.googleapis.com/...) and at runtime runs scripts/fetch-stitch.sh (curl -L "$URL") to download the remote HTML which is then injected/parsed to drive the agent's code-generation, so the external htmlCode.downloadUrl directly controls the agent output.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)