day2-create-context-sync-skill

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow (SKILL.md and templates/context-sync.md) explicitly directs Claude to fetch and ingest third‑party, user‑generated content (Slack, Gmail, Notion, public RSS/web pages) via MCP/subagent calls and to act on those results (e.g., produce highlights, action items, send Slack/Notion updates), and the included script scripts/mcp_servers.py even downloads a public GitHub README (README_URL) to discover MCP servers — meaning untrusted external content is read, interpreted, and can materially influence subsequent tool use and decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's scripts/mcp_servers.py downloads and parses the external README at runtime from https://raw.githubusercontent.com/modelcontextprotocol/servers/main/README.md and uses the fetched entries to generate installation commands and .mcp.json configuration (which directly control the agent's presented prompts/instructions and can lead to executing remote packages like via npx), so the external content meaningfully controls runtime behavior.