web3js-v1-to-v3-migration
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user code to perform migrations, creating a surface for potential indirect prompt injection attacks.
- Ingestion points: External source code files containing legacy Solana web3.js implementation (SKILL.md).
- Boundary markers: No specific delimiters are mandated for the agent when ingesting external code files.
- Capability inventory: The skill assumes the ability to modify source files and interact with Solana RPC nodes for fetching account information (reference/spl-token.md).
- Sanitization: The instructions do not include steps for the agent to sanitize or detect malicious payloads within the analyzed code.
Audit Metadata