mem0-test-integration

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly WebFetches public third-party URLs (see "Canonical sources" in SKILL.md such as https://docs.mem0.ai/llms.txt, https://raw.githubusercontent.com/mem0ai/.../SKILL.md and GitHub repo URLs) and then parses those documents to determine import signatures, test shapes, and delegated-skill example blocks that directly drive static checks, smoke tests, and E2E behavior, so externally hosted content can materially influence its actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly WebFetches canonical skill/docs at runtime (e.g. https://raw.githubusercontent.com/mem0ai/mem0/main/skills/mem0/SKILL.md) and uses the fetched example blocks as the authoritative shapes for static checks and the smoke/E2E test flow, so remote content directly controls the agent's test prompts/instructions.