agentskills-specification

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

• Third-party content exposure detected (high risk: 0.70). This skill explicitly directs the agent to read public, third‑party documentation (e.g., agentskills.io/llms.txt and the github.com/agentskills/agentskills repo) as the canonical sources for the specification, so the agent would fetch and interpret external website content that could materially affect behavior.