icp-persona-builder

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to utilize WebFetch and WebSearch tools to gather data from various third-party domains, including company websites, review platforms (such as G2, Capterra, and TrustRadius), and job boards. These external references are used as part of the primary research function to identify customer segments and motivations.
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it ingests and processes untrusted data from the web.
  • Ingestion points: Data enters the agent context through WebFetch and WebSearch outputs during the Phase 1 research process in SKILL.md.
  • Boundary markers: The instructions do not provide explicit delimiters or instructions to treat external research data as data rather than instructions, which could allow malicious content on researched sites to influence agent behavior.
  • Capability inventory: The agent has the capability to search the web, fetch web content, and write structured JSON and Markdown files to the local clients/ directory.
  • Sanitization: There are no defined procedures for sanitizing, filtering, or validating the content retrieved from external sites before it is incorporated into the persona assets.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:05 PM
Security Audit — agent-trust-hub — icp-persona-builder