The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill relies on npx agentcash@latest to perform its core functions. This pattern downloads and executes code from the npm registry at runtime without version pinning. This creates a supply chain risk, as the @latest tag ensures that any update to the package—including potentially malicious code—would be executed by the agent without verification.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process and follow endpoint-specific guidance contained in an instructions field returned by an external API call.
Ingestion points: The instructions field in the JSON output from the npx agentcash@latest discover <origin> command.
Boundary markers: Absent. The agent is explicitly told to "Read the output carefully" and follow the provided guidance without any instruction to ignore or sanitize embedded commands.
Capability inventory: The skill provides the agent with access to shell command execution (via npx) and network request capabilities (via fetch) across various external domains.
Sanitization: None identified; instructions from the external source are integrated directly into the agent's workflow.
[COMMAND_EXECUTION]: The skill utilizes several shell commands (npx, fetch, balance, discover, check) to interact with the service provider's CLI tool and perform operations like wallet management and API calls.
[EXTERNAL_DOWNLOADS]: The skill dynamically fetches the agentcash package from the public npm registry during execution.

agentcash