inttegro-sell-by-link
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external tool outputs (e.g., product lists and price details from the Inttegro MCP server) which could potentially contain malicious instructions. However, the risk is mitigated by strong instructional guardrails.
- Ingestion points: Data enters the context via
list_products,get_product,list_prices,get_price, andlist_buy_links(SKILL.md). - Boundary markers: The instructions lack explicit technical delimiters (like XML tags) but enforce a 'human-in-the-loop' boundary by requiring 'explicit merchant confirmation' for all impactful actions.
- Capability inventory: The skill can perform write/action operations including
create_product,update_product,create_product_price,update_price,create_buy_link,update_buy_link,cancel_buy_link, andsend_buy_link_sms(SKILL.md). - Sanitization: The skill explicitly forbids passing raw phone numbers and instructs the agent to accept only canonical URLs returned by the server, reducing the risk of data poisoning.
- [EXTERNAL_DOWNLOADS]: The skill configuration connects to an external Model Context Protocol (MCP) server at
https://mcp.inttegro.com/(agents/openai.yaml). This is a vendor-owned resource used for the primary functionality of the skill. - [SAFE]: The skill follows security best practices for payment handling by explicitly directing the agent to use external hosted checkouts rather than collecting sensitive payment information within the chat interface.
Audit Metadata