zhang-yiming-perspective

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to use web search (Step 2 of the workflow) to gather real-time data on products, markets, and organizations. This creates a vulnerability where malicious content on third-party websites could influence the agent's behavior.\n
  • Ingestion points: Web search results are ingested into the context via the 'Agentic Protocol' in SKILL.md.\n
  • Boundary markers: The instructions lack delimiters or specific warnings to ignore instructions embedded in search results.\n
  • Capability inventory: The skill uses git for version checks and WebSearch for information gathering (defined in SKILL.md). It also requests various file-read operations.\n
  • Sanitization: There are no mentioned mechanisms for sanitizing or escaping the data fetched from the web.\n- [COMMAND_EXECUTION]: The 'Version self-check' section in SKILL.md instructs the agent to execute shell commands (git rev-parse and git ls-remote) to compare local and remote versions. This represents intentional command execution by the skill logic.\n- [EXTERNAL_DOWNLOADS]: The README.md file recommends installing the skill using npx, which downloads and runs code from the NPM registry. It also references nuwa-skill, another tool from the same author.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 02:01 AM