The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from CLI outputs and user-provided inputs to construct and execute subsequent shell commands.
Ingestion points: Data is ingested from the output of circle wallet list, circle wallet balance, and circle gateway balance (SKILL.md, references/fiat-on-ramp.md, references/gateway-eco-deposit.md).
Capability inventory: The skill possesses extensive command execution capabilities via the Bash tool, including circle wallet fund, circle gateway deposit, and circle bridge transfer (SKILL.md).
Boundary markers: The instructions mandate that the agent must confirm the funding method and amount with the user before execution, which serves as a manual checkpoint (SKILL.md).
Sanitization: The skill depends on the internal argument parsing and validation of the circle CLI to sanitize inputs interpolated into commands.
[COMMAND_EXECUTION]: The skill's primary functionality is implemented through the execution of shell commands using the circle CLI tool.
Evidence: The allowed-tools configuration in SKILL.md explicitly permits various subcommands of the circle CLI.
Scope: Execution is scoped to the circle binary, which is used for wallet status, balance checks, funding, and transfers.

fund-agent-wallet