coreweave-hello-world

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes kubectl commands through the Bash tool to manage and deploy resources on CoreWeave clusters, which is central to its stated purpose.
  • [EXTERNAL_DOWNLOADS]: References official, well-known container images from vllm and pytorch and provides links to official CoreWeave documentation and GitHub repositories.
  • [INDIRECT_PROMPT_INJECTION]: Identifies the attack surface where user input is processed to trigger deployment tasks.
  • Ingestion points: User trigger phrases specified in SKILL.md (e.g., "run on coreweave").
  • Boundary markers: No specific delimiters or instructions to ignore embedded content are present in the provided instructions.
  • Capability inventory: The skill has access to shell execution (kubectl) and file manipulation tools (Write, Edit).
  • Sanitization: No explicit sanitization or validation of user-provided input is performed before it is used to interact with the cluster.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:43 AM
Security Audit — agent-trust-hub — coreweave-hello-world