skills/base/skills/vibenet/Gen Agent Trust Hub

vibenet

Fail

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The script scripts/setup-viem-8130.sh performs a 'clone-and-build' operation from a personal GitHub repository (https://github.com/chunter-cb/viem). This process involves downloading unverified code and executing build scripts via pnpm install --ignore-scripts and pnpm run build on the host machine. This pattern bypasses standard security reviews associated with official package registries and introduces significant supply-chain risk.
  • [COMMAND_EXECUTION]: The skill relies on an automated shell script (scripts/setup-viem-8130.sh) to modify the local development environment. The script executes multiple high-risk commands, including git clone, pnpm build, and npm install --install-links, which modify the project's dependency tree and local filesystem.
  • [EXTERNAL_DOWNLOADS]: The skill initiates network requests to several external and unverified sources:
  • Downloads repository content from https://github.com/chunter-cb/viem.
  • Interacts with RPC and API endpoints at rpc.vibes.base.org and api.vibes.base.org.
  • References external walk-throughs on gist.github.com/chunter-cb.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from blockchain RPC providers and potentially user-provided application code. While it includes some guardrails like read-back verification, it lacks explicit sanitization for data ingested from the vibenet devnet, which could be exploited if an attacker controls the RPC responses.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 14, 2026, 05:11 PM
Security Audit — agent-trust-hub — vibenet