memos-cloud

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow runs scripts/upgrade.py, which fetches remote resources/meta.json and downloads updated resources/*.md from a public GitHub raw URL; those downloaded Markdown files are then read/used by the agent (e.g., resources/cn/index.md, resources/en/index.md, starter-pack.md).

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

• Potentially malicious external URL detected (high risk: 1.00). The scripts/upgrade.py is invoked at runtime (Required Workflow step 1) and fetches and overwrites local skill resources from the remote BASE_URL (https://raw.githubusercontent.com/MemTensor/MemOS-Cloud-Skill/main/memos-cloud/resources, e.g. VERSION and meta.json and listed files), and those fetched markdown resources are then read to drive prompt routing and integration behavior—i.e., remote content directly controls agent instructions at runtime.