newsletter-curation

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external data from search tools, which is a common attack surface for indirect prompt injection.
  • Ingestion points: External content is fetched using tavily/search-assistant and exa/search based on user queries.
  • Boundary markers: None explicitly defined in the templates to separate search results from instructions.
  • Capability inventory: The skill is restricted to using the belt CLI tool via allowed-tools: Bash(belt *).
  • Sanitization: Relies on the host agent's internal safety filters when processing the retrieved web content.
  • [COMMAND_EXECUTION]: The skill uses the belt CLI to interface with remote applications for search, image generation, and social media posting. The execution environment is limited to the belt binary through frontmatter configuration.
  • [EXTERNAL_DOWNLOADS]: The skill references and provides instructions for installing the belt-sh/cli and additional skills from the inference-sh repository. These are standard dependencies for the platform the skill is intended to operate on.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:53 PM
Security Audit — agent-trust-hub — newsletter-curation