web-research

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its web research workflow.
  • Ingestion points: Untrusted data from the internet is gathered via web_search and fetch_url tools, which is then stored in local files (research_[topic_name]/findings_*.md) as specified in SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions directing the agent to ignore potentially malicious embedded commands within the fetched content before synthesis.
  • Capability inventory: The skill possesses capabilities to read/write local files (read_file, write_file) and execute subtasks through the task tool, which could be abused if malicious data influences the agent's logic.
  • Sanitization: The process lacks validation or filtering of retrieved web data before the agent reads and synthesizes it into a final report.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 03:11 AM
Security Audit — agent-trust-hub — web-research