The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the installation of an MCP server using npx -y @ruvnet/rvagent, which downloads and executes code from the NPM registry. This package is part of the vendor's own namespace and is consistent with the skill's stated purpose.
[COMMAND_EXECUTION]: Includes commands to check the published version of the vendor's package (npm view @ruvnet/rvagent version) and to execute the server via NPX.
[PROMPT_INJECTION]: The skill implements tools such as bfld_subscribe and vitals_fetch that ingest external sensing data (BFLD events, heart rates, breathing rates) into the agent's context. This creates a surface for indirect prompt injection where malicious instructions could be embedded in the sensing data stream.
Ingestion points: bfld_subscribe, vitals_get_all, and vitals_fetch tools defined in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded text within the sensing data are defined in the skill body.
Capability inventory: The agent is intended to "reason about and respond to" sensing data, with potential actions including cross-cog coordination and "paging the carer."
Sanitization: The skill references a "sanitizer" middleware component located in vendor/ruvector/crates/rvAgent/rvagent-middleware/, indicating some internal processing of event data.

ruview-rvagent