The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the robomotion CLI for task automation, including flow validation, execution, and log retrieval. These commands are specific to the vendor's own ecosystem and infrastructure.
[COMMAND_EXECUTION]: The skill provides platform-specific commands such as xdg-open, open, and start for the agent to interact with files generated by automation flows. It includes explicit guardrails requiring user consent before opening files and restricting actions to verified output paths to prevent automated execution of untrusted files.
[INDIRECT_PROMPT_INJECTION]: The skill implements a feedback loop by reading robot session logs, which are external data sources that could potentially contain untrusted content.
Ingestion points: Session logs located in platform-specific directories like ~/.config/robomotion/ or %LOCALAPPDATA%\Robomotion\ are read using robomotion run and robomotion logs in SKILL.md.
Boundary markers: No explicit delimiters are used for the log content, although events follow a structured JSONL format which provides some implicit separation.
Capability inventory: The agent has the capability to execute the robomotion CLI and system openers as defined in SKILL.md.
Sanitization: The instructions include a recommendation to verify flow success and specific file paths before performing any downstream actions based on the log output.

running-flow