Security audit of Prism-Shadow/AgentHub · Agent Tool by Prism-Shadow · ★ 72
Use with caution — AgentHub triggered warning flags in AgentSkillsHub's security scan. It may be fine for personal trials, but review its credential handling and maintainer before brand or production use.
What it is: AgentHub is the LLM API Hub for the Agent era, built for high-precision autonomous agents. (GPT-5.5/Claude 4.6/Gemini 3.1)
env access| Security grade | ⚠ CAUTION |
| Quality score | 70/100 |
| GitHub stars | 72 |
| Language | Python |
| License | Apache-2.0 |
| Last updated |
This is AgentSkillsHub's free basic audit: an automated rule-based scan covering SlowMist's 11 red-flag categories (credential exfiltration, obfuscated payloads, sandbox escape, prompt injection, and more) across 117,000+ open-source AI agent skills and MCP servers, refreshed every 8 hours. A CAUTION grade is a scan result, not a guarantee — deep 5-dimension audits (code · credentials · vendor · supply-chain · operational) are available for enterprise. Audited: 2026-07-03.