marketing-video

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow reads user-provided/inter-tenant free text from the user via the “Interview” conversation (e.g., product description, audience, desired action, storytelling/taste, deadlines, and asset descriptions) before producing the creative brief and subsequent steps.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs a runtime installation "npx skills add remotion-dev/skills" which pulls the external code from https://github.com/remotion-dev/skills and installs 38 rule modules into ~/.claude/skills/ that directly influence the agent's behavior and prompts.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly references crypto data and blockchain libraries: it directs pulling live TVL/volume via DefiLlama and CoinGecko (crypto data APIs) and explicitly names @solana/web3.js for pulling live on-chain data. These are specific blockchain/crypto APIs (not generic tools) and fall under the crypto/blockchain category in the policy, so the skill contains explicit crypto-related execution capabilities (and thus should be flagged).