skills/npc-live/clawfirm/func/Gen Agent Trust Hub

func

Warn

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to execute a binary at ~/projects/pi-go/bin/func. The commands are constructed using formulas and file paths that can be influenced by user input.
  • [EXTERNAL_DOWNLOADS]: The instructions explicitly suggest using curl to download data if it is not provided by the user, which involves making network requests to arbitrary external domains.
  • [REMOTE_CODE_EXECUTION]: The execution flow lacks sanitization logic when interpolating user-provided formulas or file paths into the shell command ~/projects/pi-go/bin/func '<公式>' <csv文件路径>. This creates a vulnerability to command injection where shell metacharacters could be used to execute arbitrary instructions.
  • [DATA_EXFILTRATION]: The combination of file system access (reading CSV files) and network capabilities (via curl) provides a potential path for reading sensitive data and transmitting it to an external server.
  • [EXTERNAL_DOWNLOADS]: The skill references cdnjs.cloudflare.com to load the ECharts library for visualization; this is a well-known service and is considered a safe resource reference.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 20, 2026, 06:07 AM
Security Audit — agent-trust-hub — func