coreweave-security-basics

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for ingesting and processing untrusted data from external webhooks and user-defined workload requests, creating a potential attack surface.
  • Ingestion points: Processes webhook payloads in verifyCoreWeaveWebhook and user-provided workload data in validateWorkloadRequest.
  • Boundary markers: Recommends the use of cryptographic signatures (HMAC) and schema-based validation (Zod) to ensure data integrity and prevent injection.
  • Capability inventory: Utilizes kubectl via the Bash tool for namespace and infrastructure management, along with standard file system operations.
  • Sanitization: Provides robust code for input sanitization using regular expressions and logic for redacting sensitive fields from logs.
  • [COMMAND_EXECUTION]: The skill provides instructions for executing kubectl commands to apply network policies and run diagnostic checks within a Kubernetes cluster, which is consistent with its stated purpose of cloud security management.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:44 AM
Security Audit — agent-trust-hub — coreweave-security-basics