The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation encourages the use of shell commands (curl) to interact with the Sponge Wallet API. This includes executing high-privilege operations such as transferring tokens across chains (EVM, Solana, Tempo), performing swaps, and initiating bank off-ramps. These actions represent significant capabilities that could be misused if the agent's logic is subverted.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its integration with 'paid external services' like web scraping and search. Malicious instructions embedded in external web content processed by these services could attempt to influence the agent's behavior.
Ingestion points: Untrusted data enters the agent context through endpoints like /api/paid/fetch, /api/discover, and /api/checkout (online store URL processing).
Boundary markers: No specific delimiters or safety instructions are provided to help the agent distinguish between its core instructions and the potentially adversarial content returned from these external sources.
Capability inventory: The skill provides the agent with extensive financial capabilities, including POST /api/transfers/*, POST /api/transactions/*, and POST /api/bank/send.
Sanitization: There is no documentation regarding the sanitization or validation of the data retrieved from external URLs before it is processed by the AI agent.

sponge-wallet