The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use a curl command to register the automation with the platform's backend. This operation uses environment variables provided by the agent's runtime environment to communicate with the vendor's own infrastructure.
[DATA_EXFILTRATION]: The skill transmits the research prompt and schedule to the OPENHANDS_HOST. As this host is part of the vendor's official automation infrastructure, the data transfer is considered a standard operational requirement for the skill's stated purpose.
[INDIRECT_PROMPT_INJECTION]: The skill summarizes content gathered from the web via the Tavily MCP. This creates a potential surface for indirect prompt injection if malicious instructions are encountered on external websites. However, this is an inherent risk for research-oriented tasks, and no evidence of targeted exploitation is present in the skill code.

research-brief