competitive-teardown

Pass

Audited by Gen Agent Trust Hub on Apr 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is focused on product and business research. No security findings related to credential theft, data exfiltration, or malicious command execution were detected. The use of well-known professional platforms (LinkedIn, G2, Reddit) as information sources is standard for this task category.
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection due to its core function of processing untrusted web content from competitor sites.
  • Ingestion points: External data retrieved through the WebSearch and WebFetch tools as specified in SKILL.md.
  • Boundary markers: The skill does not provide explicit delimiters or "ignore previous instructions" safety wrappers for the fetched external content.
  • Capability inventory: Capabilities are limited to WebSearch, WebFetch, and Read (of local templates); no high-risk capabilities like arbitrary shell execution or system-wide file writes are requested.
  • Sanitization: No specific validation or filtering of external content is defined in the instructions, relying on the underlying agent's safety guardrails.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 9, 2026, 01:07 PM
Security Audit — agent-trust-hub — competitive-teardown