agentic-wallet

Pass

Audited by Gen Agent Trust Hub on Aug 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill extensively uses the npx awal@2.12.0 command to perform wallet status checks, authentication, balance queries, token transfers, and trades.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes npx to fetch the awal tool and provides instructions for installing various Node.js packages such as express and @x402/* suites for building payment servers.
  • [REMOTE_CODE_EXECUTION]: The skill downloads and executes the awal CLI tool at runtime. Additionally, the instructions for building an x402 server include writing and executing a local Node.js script.
  • [DATA_EXFILTRATION]: The x402 pay functionality allows the agent to interact with arbitrary external URLs. While intended for payment protocols, this capability permits sending data to remote endpoints.
  • [PROMPT_INJECTION]: The skill includes robust defensive instructions, requiring the agent to validate user-provided data (emails, addresses, amounts, and SQL queries) against specific regular expressions before using them in shell commands to mitigate command injection risks.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from the x402 bazaar and API responses, which represents a potential attack surface. 1. Ingestion points: Results from x402 bazaar search, x402 details, and responses from endpoints called via x402 pay. 2. Boundary markers: The skill mandates single-quoting for arguments and strict validation of inputs before command construction. 3. Capability inventory: The Bash tool is allowed for npx, npm, node, curl, and mkdir operations. 4. Sanitization: Comprehensive regex validation rules are provided for all variable inputs to prevent shell metacharacter expansion.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 5, 2026, 02:31 AM
Security Audit — agent-trust-hub — agentic-wallet