press-release-writing

Warn

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the 'belt' CLI from an external GitHub repository (inference-sh/skills) that is not included in the trusted vendor list.
  • [REMOTE_CODE_EXECUTION]: Instructions guide the agent to install additional skills from unverified third-party sources using 'npx skills add'.
  • [DATA_EXFILTRATION]: The skill performs network requests to non-whitelisted domains including inference.sh, tavily.com, and exa.ai to fetch research data.
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute 'belt' commands for searching, fact-checking, and account authentication ('belt login').
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to the ingestion of untrusted data from external search providers.
  • Ingestion points: Results from tavily/search-assistant and exa/search tools in SKILL.md.
  • Boundary markers: No delimiters or protective instructions are used for the external content.
  • Capability inventory: The skill has permission to execute shell commands via the Bash tool.
  • Sanitization: No evidence of validation or filtering of external input before it is processed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 19, 2026, 02:15 AM
Security Audit — agent-trust-hub — press-release-writing