omnigraph-intel-bootstrap

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most links are to expected, legitimate resources (Docker docs, a GitHub repo, and localhost health endpoints), but the prompt explicitly instructs fetching and executing a raw shell script from raw.githubusercontent.com (curl | bash) — a high-risk pattern because running remote scripts or arbitrary repo code can deliver malware if the source or supply chain is compromised or untrusted.

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

• Third-party content exposure detected (high risk: 1.00). The skill's Phase 6 research workflow (SKILL.md "Phase 6 — Research, seed, init, load" and references/research.md) explicitly instructs the agent to use WebFetch/WebSearch to pull recent items from public sources — including newsletters, blogs, X accounts, forums and other third‑party websites — and to extract dated signals from those pages to drive schema/seed generation, meaning untrusted web content is read and can materially influence subsequent actions and tool use.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the user at runtime to run a remote script via curl piped to bash — "curl -fsSL https://raw.githubusercontent.com/ModernRelay/omnigraph/main/scripts/local-rustfs-bootstrap.sh | bash" — which fetches and immediately executes remote code as a required bootstrap step for RustFS/omnigraph, so it is a high-confidence runtime remote-execution dependency.