writing-profile

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

• Third-party content exposure detected (high risk: 0.80). The skill's run-unit and excerpt-selection instructions require the agent to generate and present "real excerpts from real authors" (see references/excerpt-strategy.md and references/phases/run-unit.md), including modern blogs and public literary anchors (e.g., 阮一峰, CoolShell, network/blog excerpts), so the agent will ingest/display open/public third‑party text as part of its scoring workflow and those excerpts can materially influence decisions (profile scores and downstream writing behavior).