kbo-results
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute a remote CLI tool using
npx -y @nomadamas/k-skill@0to retrieve its primary instructions and supplementary files. This pattern relies on the availability and integrity of the external NPM package at runtime.\n- [COMMAND_EXECUTION]: The workflow requires the global installation of an external package (npm install -g kbo-game) and executes shell commands to run Node.js scripts via standard input.\n- [EXTERNAL_DOWNLOADS]: The skill fetches code and instructions from external registries (NPM) at runtime to perform its functions.\n- [DYNAMIC_EXECUTION]: The skill dynamically constructs a file path to import thekbo-gamemodule at runtime usingpath.join(process.env.GLOBAL_NPM_ROOT, ...)andpathToFileURL. This method of loading code is used to interface with globally installed packages.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from an external source (KBO game schedules/results via thekbo-gamepackage). While this is the intended purpose of the skill, it creates an attack surface where malicious content embedded in the sports data could potentially influence the agent if not properly sanitized.\n - Ingestion points: Output from the
kbo-gameNPM package (instruction.md).\n - Boundary markers: None specified for the data ingestion process.\n
- Capability inventory: Shell command execution (
npm,npx,node).\n - Sanitization: The instructions recommend "normalizing for humans" but do not specify technical sanitization methods for the external data.
Audit Metadata