The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines instructions for the agent to execute various commands using the x07 and x07-os-runner tools. These commands allow for running projects, passing arguments, and executing individual program files on the host operating system. While the skill encourages the use of a sandboxed profile governed by a policy file, it inherently enables direct command execution capabilities.
[PROMPT_INJECTION]: The skill instructions direct the agent to parse JSON reports produced by the x07-os-runner backend. This constitutes an indirect prompt injection surface where a malicious or compromised program could output data designed to influence the agent's logic or bypass safety constraints.
Ingestion points: The x07-os-runner.report JSON object printed to stdout during execution in SKILL.md.
Boundary markers: The output is structured as a JSON report, providing a schema-based boundary between the program output and the agent's instructions.
Capability inventory: The skill possesses extensive capabilities for file system access, network interaction, and process spawning via the x07 toolset, as evidenced by the run-os-policy.sample.json configuration.
Sanitization: The skill does not explicitly mention sanitization or validation of the content within the output field of the JSON report before the agent processes it.

x07-os-run