The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external Coinbase API endpoints, such as market data, research catalogs, and portfolio information, which could potentially contain adversarial instructions aimed at influencing the agent's behavior. * Ingestion points: Untrusted data enters the agent context through the outputs of tools like coinbase_products_list, coinbase_x402_resources, and various market data retrieval tools described in references/market-data.md. * Boundary markers: The skill provides explicit instructions in SKILL.md to treat API results and provider content as untrusted data and forbids the agent from interpreting them as instructions to modify permissions or reveal secrets. * Capability inventory: The agent has access to powerful financial tools across several categories, including order creation (coinbase_orders_create), fund transfers (coinbase_transfer), and research payments (coinbase_x402_fetch). * Sanitization: The instructions require mandatory user confirmation for all financial writes and the use of structured arguments to prevent command-style injection.
[NO_CODE]: The skill is composed entirely of markdown documentation and instructional guides. It does not ship with any executable code, scripts, or package manifests, significantly reducing the direct attack surface for traditional malware or remote code execution.

coinbase-mcp