fix-bug

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests user-generated telemetry and ticket content from third-party services (Dash0 and Linear) via MCP tools (see Phase 0 input classification in SKILL.md and the "Dash0 resolution" / "Linear input" procedures in rules/evidence-resolution.md), and that external evidence is incorporated into the Evidence Record which materially drives triage, planning, and autonomous actions—creating a clear vector for indirect prompt injection.