brief

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text is ingested during the scheduled PREP via scripts/gather.sh → mywork/sync-jira.py and mywork/slackdump which write Jira comment threads and Slack thread Markdown into ~/loop-me/state/comms/..., and then scripts/prep.sh feeds scripts/prep-prompt.md plus the staged brief composition into the claude -p LLM context (so outsider-authored Jira/Slack content can be included in the generated brief).