self-improvement

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

• Third-party content exposure detected (high risk: 1.00). The skill's "Learning from GitHub PRs and Issues" section instructs the agent to use the gh CLI to fetch and analyze GitHub PRs/issues (user-generated, public content), which the agent will read and use to generate lessons and drive follow-up actions — exposing it to untrusted third-party content that could contain indirect prompt injections.