solana-setup

Fail

Audited by Gen Agent Trust Hub on May 7, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill executes a remote installation script from the author's GitHub repository using the curl | bash pattern. While this is the intended mechanism for the 'solana-setup' skill, executing unverified remote scripts poses a security risk if the source is compromised.- [REMOTE_CODE_EXECUTION]: Version checks are performed by piping output from the Crates.io API directly into python3. Although this is used for JSON parsing, executing code based on remote content is a sensitive operation.- [EXTERNAL_DOWNLOADS]: Fetches executable shell scripts and version metadata from external domains including GitHub (github.com) and the official Rust package registry (crates.io).- [PROMPT_INJECTION]: The skill utilizes $ARGUMENTS to determine its execution flow (status, install, or update). The processing of these untrusted user inputs without explicit sanitization or strict boundary markers creates an attack surface for indirect prompt injection or command manipulation.
  • Ingestion points: $ARGUMENTS reference in SKILL.md
  • Boundary markers: Absent; user input flows directly into logic selection
  • Capability inventory: Shell execution (Bash), network operations (curl), file system queries (command -v), and dynamic code execution (python3, cargo)
  • Sanitization: Absent; the agent is instructed to "Review: $ARGUMENTS" without specific validation rules.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/widnyana/solana-onchain-mcp/refs/heads/main/install.sh, https://crates.io/api/v1/crates/solana-onchain-mcp - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
May 7, 2026, 11:21 AM
Security Audit — agent-trust-hub — solana-setup