helius-phantom

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation includes instructions for installing the Helius MCP server and various software development kits. Each of these downloads targets official repositories and registries associated with well-known Solana infrastructure providers.
  • Fetches the helius-mcp tool via npx as part of the initial setup process documented in SKILL.md.
  • References numerous official packages from the npm registry, including @phantom/react-sdk, @solana/kit, and Metaplex Foundation libraries.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted data from the Solana blockchain, which could potentially contain malicious instructions embedded by third parties.
  • Ingestion points: Processes wallet addresses, transaction signatures, and metadata retrieved via Helius DAS and Enhanced Transactions APIs (e.g., in references/helius-das.md and references/helius-enhanced-transactions.md).
  • Boundary markers: Code examples for rendering blockchain data do not include specific delimiters or instructions for the AI to ignore embedded content.
  • Capability inventory: The skill possesses network communication capabilities to interact with the Helius API and the ability to execute shell commands for project setup.
  • Sanitization: The skill explicitly includes security patterns for verifying message signatures and transaction timestamps to mitigate risks during authentication and token gating.
  • [COMMAND_EXECUTION]: The skill guides the user and agent through the use of command-line tools for environment setup and account management.
  • Documents the usage of the Helius CLI for generating wallet keypairs and performing autonomous account signup operations in references/helius-onboarding.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 03:03 PM
Security Audit — agent-trust-hub — helius-phantom