The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local script located at 'scripts/run-state-check.js' using Node.js to verify the state of the ICP process. This allows for local command execution within the agent environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted external data such as 'evidence', 'quotes', and 'market-research'. This introduces an attack surface where malicious instructions hidden in the input data could attempt to subvert the agent's logic. * Ingestion points: Market research data, customer complaints, and user quotes used to define the ICP. * Boundary markers: The instructions do not define any delimiters or isolation markers to separate external data from the system prompt or agent instructions. * Capability inventory: The skill has the capability to execute shell commands via a Node.js script. * Sanitization: No specific sanitization, validation, or escaping of the external input data is described in the skill.

icp