opensea
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external NFT and collection metadata which is user-generated and could contain malicious instructions designed to manipulate agent behavior.
- Ingestion points: External data enters the agent context through API responses fetched by core utility scripts, such as
opensea-api/scripts/opensea-get.shandopensea-post.sh. - Boundary markers: The skill explicitly implements boundary markers (
--- BEGIN OPENSEA API RESPONSE ---and--- END OPENSEA API RESPONSE ---) in its core scripts to clearly delineate untrusted API content from system instructions. - Capability inventory: The skill suite possesses extensive capabilities, including executing onchain trades (
opensea-marketplace), swapping tokens (opensea-swaps), and configuring wallet signing providers (opensea-wallet). - Sanitization: The instruction set in
opensea-api/SKILL.mdspecifically directs the agent to treat all content between the boundary markers as untrusted and never execute instructions found within them, providing a strong defense against data-borne attacks.
Audit Metadata