The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation guide in INSTALL.md for macOS and Linux uses a curl | sh pattern to execute a shell script directly from the author's GitHub repository (planetarium/a2a-x402-wallet).
[COMMAND_EXECUTION]: The skill frequently executes the a2a-wallet CLI tool to perform various tasks, including message streaming, payment signing, and authentication.
[EXTERNAL_DOWNLOADS]: The skill downloads the CLI binary and its installation script from the author's GitHub repository. It also references the x402 payment specification from Google's official agentic-commerce repository.
[DATA_EXFILTRATION]: The skill manages sensitive wallet information, including private keys (indirectly via the CLI), balances, and bearer tokens (A2A_WALLET_TOKEN). It transmits data and signed payloads to external agent URLs provided by the user or found in agent cards.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface.
Ingestion points: Processes external agent cards and response metadata provided by remote A2A agents (SKILL.md).
Boundary markers: No explicit delimiters are used to separate agent-provided data from command-line arguments.
Capability inventory: The a2a-wallet tool can sign arbitrary messages and process payments, which could be triggered by malicious agent responses.
Sanitization: There is no evidence of sanitization or validation of data received from external agents before it is used in subsequent CLI operations.

a2a-wallet