cavos-cli

What This Skill Does

The Cavos CLI skill provides an interface for OpenClaw agents to perform native Starknet operations via the @cavos/cli package. It bridges the gap between natural language requests and complex blockchain interactions, enabling agents to manage identities, perform cross-wallet transfers, interact with smart contracts, and track transaction lifecycles directly from the command line. By leveraging structured JSON output, the agent maintains high accuracy when processing blockchain data.

Installation

To integrate this skill into your environment, use the OpenClaw Hub command: clawhub install openclaw/skills/skills/adrianvrj/cavos-cli-skill

Ensure that you have Node.js installed and the necessary session tokens provisioned from the Cavos Dashboard to authenticate your agent's wallet.

Use Cases

• Automated Financial Management: Agents can monitor specific token balances (e.g., STRK or ETH) and trigger transfers to cold storage or other wallets based on pre-defined triggers.
• Smart Contract Interop: Perform complex protocol interactions such as approving ERC-20 spending allowances or executing custom contract entrypoints for decentralized applications.
• Transaction Auditing: Automate the verification of transaction statuses using hash monitoring, ensuring that every on-chain action is finalized and valid before proceeding with subsequent workflow steps.
• Batch Processing: Use the multicall feature to combine several contract operations into a single transaction, optimizing gas usage and ensuring atomicity.

Example Prompts

1. "Check my current ETH balance and if it's over 0.5, send 0.1 ETH to 0x047..."
2. "Execute a multicall to approve 50 STRK for the exchange contract and then swap them for USDC."
3. "What is the status of transaction hash 0x3a... and who is the current session wallet holder?"

Tips & Limitations

• Always Authenticate: Start workflows by confirming your session status. If a session is expired, the agent must be prompted to re-import the token via the Dashboard.
• Data Integrity: Always use the --json flag to avoid parsing errors. Relying on unstructured CLI output can lead to failures during complex agent reasoning.
• Calldata Formatting: Ensure input for calldata is consistently comma-separated (e.g., '0x123,456'). Improperly formatted arguments will fail to parse at the contract level.
• Simulations First: Always run the simulate command before executing high-value transactions to ensure the transaction will succeed on-chain.