skills/eumetsat/skills/ewccli/Gen Agent Trust Hub

ewccli

Pass

Audited by Gen Agent Trust Hub on Oct 24, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements strong security practices by explicitly forbidding the display or logging of sensitive credentials, such as OpenStack application secrets and private SSH keys. It uses placeholders in all command examples and provides scripts to validate local configurations without exposing secret values. All referenced domains (e.g., europeanweather.cloud, ecmwf.int, cloudferro.com) and the primary tool (ewccli) are official resources for the European Weather Cloud infrastructure.
  • [INDIRECT_PROMPT_INJECTION]: The skill supports an attack surface by allowing the processing of user-provided YAML files via the --path-to-catalog flag for testing custom EWC Community Hub items.
  • Ingestion points: The ewc hub command reads external YAML files specified by the user as documented in references/custom-catalogues.md and references/hub.md.
  • Boundary markers: Rule 3 in SKILL.md requires explicit user confirmation before executing state-changing commands, and the response contract in SKILL.md mandates credential safety and validation status reporting to the user.
  • Capability inventory: The skill uses the ewc CLI to list, show, and deploy infrastructure, which includes creating OpenStack servers and running Ansible playbooks as described in references/infra.md and references/hub.md.
  • Sanitization: The skill advises users to parse the YAML with a dedicated parser and run dry-run deployments (--dry-run) to validate item metadata and resource impact before live execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 24, 2026, 12:24 PM