competitive-intel
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze data from external, untrusted sources, which serves as a vector for indirect prompt injection.\n
- Ingestion points: The skill references sources such as competitor websites, product changelogs, customer review sites (G2, Capterra, Trustpilot), and social media platforms (LinkedIn, Twitter/X, Facebook Ad Library) as described in
references/ci-playbook.mdandSKILL.md.\n - Capability inventory: While not explicitly containing code, the instructions assume the agent will use network-enabled tools to fetch and process this external data.\n
- Boundary markers: There are no instructions to use delimiters or ignore embedded prompts when processing retrieved external content.\n
- Sanitization: No specific data validation or filtering steps are provided to mitigate risks from external data ingestion.
Audit Metadata