laravel-starter-kit-upgrade

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

• Third-party content exposure detected (medium risk: 0.30). In Phase 2 and Phase 3 the workflow uses gh api / gh pr list to read outsider-authored GitHub commit/PR free text (commit messages and PR titles) from laravel/<kit> and feeds it into the feature catalog and selection loop, before any specific repo item is selected by the user.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill fetches upstream kit data at runtime (used via kit_dir=$(scripts/fetch_kit.sh ) and by gh api calls) from https://github.com/laravel/$kit.git (and the GH API endpoint repos/laravel//commits?sha=&per_page=100), and that fetched content is required and used to drive the agent's feature catalog and the files it writes into the user's repo, so it directly controls the agent's runtime behavior.