recall

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

• Third-party content exposure detected (medium risk: 0.65). SKILL.md instructs the agent to read the workspace conversation transcript / archive and other workspace artifacts (e.g., shared record, tickets, PRs, issues) and then “write the brief,” so the LLM runtime context can include outsider-authored text from those sources (e.g., issue/PR/ticket body authored by non-operating user or external collaborators).