midnight-compact

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

• Third-party content exposure detected (high risk: 0.70). The skill's required CLI workflow in SKILL.md instructs fetching and running public resources (e.g., the compact-installer.sh via curl from github.com and recommending the Midnight faucet and multiple community GitHub repos), which are open/public third-party sources whose content would be ingested as part of the documented workflow and could therefore influence tool use or actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill documentation instructs users to run a shell pipeline that fetches and executes remote code from https://github.com/midnightntwrk/compact/releases/latest/download/compact-installer.sh (curl ... | sh), which downloads and runs installer code at runtime and is presented as a required toolchain installation—this is a high-confidence remote-code-execution risk.

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

• Direct money access detected (high risk: 1.00). The skill explicitly exposes blockchain coin operations and wallet integrations that can move or create value. Examples in the prompt include coin-specific APIs like receive(coin), sendImmediate(coin, recipient, amount), mintShieldedToken(...), and shielded variants (receiveShielded / sendImmediateShielded). It also references wallet connectivity types (ZswapCoinPublicKey, CoinInfo), SDKs that handle fee payment (DustWallet), and deployed examples for token minting and token swaps. These are specific crypto/blockchain transaction functions (send/mint/receive) — i.e., direct financial execution capabilities.