cournot
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes an external command-line tool to handle wallet operations and signatures.
- The
scripts/cournot-client.mjsfile usesspawnSyncto invoke a wallet CLI (defaulting tobaw, the Binance Agentic Wallet). - It includes a
walletInvocationhelper that manages command construction and argument escaping for both Unix-like and Windows environments, utilizingcmd.exe /con Windows with specific character escaping. - [INDIRECT_PROMPT_INJECTION]: The skill processes structured data and natural language evidence (basis data) from the Cournot API, which represents an attack surface for indirect prompt injection.
- Ingestion points: Data enters the agent's context through API responses from
https://interface.cournot.aiandhttps://dev-interface.cournot.ai, processed inscripts/account-flow.mjsand formatted inreferences/response-format.md. - Boundary markers: The
references/response-format.mdfile provides explicit instructions to treat API results as authoritative and prevents the agent from supplementing the output with external or invented information. It also mandates escaping markdown delimiters like|when rendering tables. - Capability inventory: The skill has the ability to execute CLI commands (
spawnSync), perform network requests (fetch), and write to the local filesystem (writeFileSyncfor credentials and payment intents). - Sanitization: The
scripts/account-flow.mjsfile includes asanitizefunction that redacts sensitive fields (such asapi_key,signature, andprivateKey) from tool outputs and logs. - [EXTERNAL_DOWNLOADS]: The skill documentation includes instructions for installing a trusted external dependency.
references/payment.mdprovides a command to install thebinance-agentic-walletskill from the official Binance skills repository using the platform's skill management tool.
Audit Metadata