sglang-skill

Warn

Audited by Snyk on Mar 2, 2026

Risk Level: MEDIUM
Full Analysis

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

  • Third-party content exposure detected (high risk: 0.80). The skill explicitly clones and reads a public GitHub repository (update-sglang.sh pulls https://github.com/sgl-project/sglang.git into the local SGLANG_REPO and SKILL.md directs the agent to grep and inspect files under that repo), which is untrusted third-party content the agent is expected to interpret and that could materially influence decisions.

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

  • Potentially malicious external URL detected (high risk: 1.00). The skill's updater script fetches and updates code at runtime from https://github.com/sgl-project/sglang.git, and that repository contains executable Python/C++/CUDA code which the skill runs (e.g., python -m sglang.launch_server), so the fetched content can execute remote code.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 2, 2026, 03:51 AM