pump-ai-agents
Warn
Audited by Snyk on Jun 21, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 0.85). The MCP server entry ("command": "npx", "args": ["-y", "@pump-fun/mcp-server"]) will fetch and execute remote code from the npm registry at runtime (e.g. https://registry.npmjs.org/@pump-fun/mcp-server) and the skill configuration requires that runtime component for agent tool integration, so it directly executes external code.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 1.00). The skill documentation explicitly references blockchain wallet and transaction tooling: it mandates Solana crypto libraries (solana-sdk, @solana/web3.js, solana-keygen), enforces keypair handling and zeroization, references createInstruction/createV2Instruction (on-chain instructions), and registers an MCP server named "solana-wallet" (npx @pump-fun/mcp-server) for stdio transport. These are specific crypto/wallet integrations and transaction-related APIs — not generic tooling — and therefore enable direct financial execution (signing/sending blockchain transactions).
Issues (2)
W012
MEDIUMUnverifiable external dependency detected (runtime URL that controls agent).
W009
MEDIUMDirect money access capability detected (payment gateways, crypto, banking).
Audit Metadata