fabric-cli

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to "smart-save" and operate on arbitrary URLs (e.g., SKILL.md and references/command-reference.md show examples like fabric --json save "https://example.com"), and later workflows (references/workflows.md, references/agent-memory.md) have the agent search, ask, summarise, or use saved web content as persistent memory—meaning untrusted public web pages can be fetched and then read/interpreted to influence subsequent actions.