Template Discovery

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow (SKILL.md and related docs) mandates WebFetch of public GitLab raw URLs (e.g., https://gitlab.com/to-be-continuous/[name]/-/raw/master/README.md and template YAML files) and the agent must parse those user-maintained repository files to decide variants and next actions, so publicly-hosted, user-generated content can influence tool use and behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs runtime WebFetch calls to raw GitLab URLs (e.g., https://gitlab.com/to-be-continuous/[template-name]/-/raw/master/README.md and similar template YAML links) and injects that fetched content into the agent's analysis/prompting, so external content directly controls agent behavior at runtime.