The Agent Skills Directory

[DATA_EXFILTRATION]: The skill facilitates the creation of public web shares from local sandbox ports using the islo surface web command. If an agent misidentifies internal services (such as databases, administrative panels, or debug ports) as user-facing interfaces, it may unintentionally expose sensitive data to the public internet.- [COMMAND_EXECUTION]: The islo surface terminal command allows for the execution of arbitrary shell commands as part of a persistent terminal session. This capability provides a direct mechanism for code execution within the sandbox environment.- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to instructions embedded in the data it processes, which could manipulate the agent into misusing its publishing capabilities.
Ingestion points: The agent parses the project repository and the running service stack to select content for publication (references/usage.md).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when the agent evaluates the untrusted content in the repository.
Capability inventory: The skill possesses the ability to create public network shares and execute arbitrary shell commands.
Sanitization: While the instructions provide guidelines for the agent to manually check for secrets and private data, there is no technical sanitization or validation of the interfaces being exposed.

surfaces