memory-view

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

• Third-party content exposure detected (high risk: 0.85). The required workflow runs scripts/scan-memory.ts which reads ~/.claude/projects/<slug>/memory/MEMORY.md and topic markdown bodies (outsider-authored user content from disk) and then injects that derived text into the model layer via serializeFindings(...) JSON printed to stdout (args.mode === "audit"), which the agent consumes as LLM context.