khe-brain-isb

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

• Third-party content exposure detected (low risk: 0.10). The required workflow reads a pinned, bundled corpus (the isb-skills rule files and the ivi session notes), and the only runtime fetch is the optional Refresh step against the public HKTITAN/isb-skills repository, which the author controls and which is not a queue an outsider can post into, so the exposure is low rather than medium.

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

• Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is the setup install commands, which instruct the user to fetch installable skill material from HKTITAN/isb-skills and HKTITAN/skills, both individual-owned GitHub repositories (unofficial, common/reputable at best), while all cited GitHub commit URLs are static citations and the notes are retrieved data.