superbuild

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly requires ingesting a user-provided plan document (Step 1: "Ingest Plan") and then parses directives from that plan (e.g., CHECKPOINT parsing and auto-compact in Step 9 and related sections) which the agent is expected to read and act on, so arbitrary/untrusted plan content can influence execution and tool use.