continuous-learning-v2

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

• Third-party content exposure detected (high risk: 0.90). The skill's CLI explicitly fetches and ingests arbitrary HTTP(S) URLs in scripts/instinct-cli.py (cmd_import uses urllib.request.urlopen to fetch a provided URL) and then parses and imports that untrusted third‑party content into the project's/global instincts which are later loaded and used to drive evolve/promote behaviors and generate skills/agents, allowing external content to materially influence agent actions.