strategy-builder

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

• Third-party content exposure detected (medium risk: 0.30). The workflow (“strategy-builder”) ingests outsider-authored free text via the operator-provided ${var} (especially goal=<free text> in the dashboard brief), and it also runs WebFetch on user-supplied links= values to summarize the fetched pages.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill fetches external repo READMEs via the gh api and WebFetches arbitrary links at runtime (e.g., the example "repo=acme/widgets" and "https://acme.com"), and that fetched content is used as contextual input to generate the agent's STRATEGY (i.e., it directly influences prompts/output at runtime).