The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to download and install the 'ACP CLI' from an external repository at https://github.com/Virtual-Protocol/openclaw-acp.
[COMMAND_EXECUTION]: The defined workflow relies on executing a series of shell commands using the downloaded acp tool, including acp setup, acp browse, and acp job create.
[COMMAND_EXECUTION]: There is a risk of command injection as user-provided data (contract_address) is interpolated directly into a shell command template (acp job create ... --requirements '{"contract_address": "..."}').
Ingestion points: The contract_address parameter enters the agent context through user input as defined in SKILL.md.
Boundary markers: No delimiters or instructions to ignore embedded malicious commands are present in the command template.
Capability inventory: The skill documentation explicitly requires the ability to execute shell commands (subprocess calls) to interact with the acp CLI.
Sanitization: No input validation or escaping mechanisms are specified to ensure the contract_address does not contain shell metacharacters.
[NO_CODE]: The skill package contains no executable scripts (e.g., .py, .js, .sh), consisting only of markdown instructions and JSON evaluation files.

solana-token-verification