CLAUDE CODE AGENTS CAN NOW EXECUTE ONCHAIN 🤯
@agentlayer_ai just shipped AgentLayer support for Claude Code, bringing wallets, DeFi, and onchain execution directly into the agent workflow.
Inside Claude Code:
- Send transfers
- Execute swaps
- Access DeFi
- Operate across Solana and EVM
- Hold and manage assets
AgentLayer just gave Claude Code agents a wallet, and the line between coding agents and economic agents is disappearing fast.
Claude Code can now run a wallet
shipped the AgentLayer plugin: transfers, swaps, and DeFi across EVM and Solana - from inside the agent
this significantly expands what the agent can do: hold value, pay, and act onchain right inside the claude flow
vibecoding now reaches money and onchain actions, not just code
local-first, fully keys control on your side
AGENTLAYER JUST PLUGGED CODEX INTO THE ONCHAIN ECONOMY 🤯
@agentlayer_ai is bringing wallets, DeFi, and machine-native payments directly into Codex.
Inside Codex, agents can now:
- Use a multichain wallet
- Access Solana, EVM, and Bitcoin
- Run DeFi directly from the agent framework
- Discover and pay for services through x402
- Swap and bridge assets across chains
Through x402, agents can autonomously purchase capabilities like:
- Web data extraction
- Image generation
- Speech-to-text
- OCR
- Browser automation
- LinkedIn intelligence
→ Pay per request → No subscriptions → No manual checkout flows
AgentLayer is turning Codex from a coding environment into an economic environment for AI agents.
agentLayer wallet now runs inside codex
your existing local AgentLayer wallet: Solana, EVM, Bitcoin works natively in the coding agent
- discover x402 providers that extend your agent’s capabilities
- swap and bridge across chains
- provide your agent ability to run DeFi
secure by design - keys never leave your machine
Agent Wallet 0.1.34 is live
- Agents can now run @VeloraDEX a swaps with native ETH,no wrapping, no extra hop
- Your agent knows when it's out of date: a background check surfaces a notice once per session, never blocks startup, fails silent offline
- Context optimisation for Solana tools
Less version archaeology. More execution.
already available
Most of my time building AgentLayer over the past week has been dedicated to reliability and security
The challenge is balancing those priorities with another major constraint in agentic systems: the context window
Building a good plugin (app) for openclaw/codex, and similar platforms is only half the job. The other half is minimizing token consumption
If your plugin becomes too expensive in terms of context usage, people will eventually stop using it, no matter how useful it is
That’s a new standard for AI apps