1inch is pushing deeper into AI-native finance with an update to its MCP Server that gives AI agents direct access to its developer ecosystem, including swap-related workflows and broader API tooling. The company says the server connects coding assistants and agents to 1inch documentation, API references, and production-ready SDK examples, so users can search, retrieve examples, and build integrations without jumping between tabs or manually copying code. According to the company’s docs, the MCP Server is hosted at api.1inch.com/mcp/protocol and there is no separate charge simply to connect to it.
The bigger shift is that this is no longer just a documentation helper. With authentication, the MCP layer can support swap execution and other onchain actions through the same business and developer stack used by the rest of 1inch’s APIs. The docs say the server exposes tools for documentation search, example discovery, example source retrieval, swaps, orderbook actions, and API queries, while authenticated calls can build quotes, execution steps, and trading workflows across classic, intent-based, and cross-chain swaps.
That makes the update especially relevant as more builders experiment with autonomous agents that can research, decide, and act inside a single loop. 1inch says the idea is to let AI systems plan and execute workflows using live infrastructure rather than stale snippets or disconnected endpoints. The company’s portal also describes a broad API surface that spans portfolio, balance, gas price, spot price, token, NFT, transaction gateway, charts, domains, and other endpoints, giving agents the data context they need before any trade or onchain action is attempted.
Making Web3 Autonomous for Growth
Still, 1inch is not presenting this as unsupervised automation. Its docs make clear that authentication, entitlements, and transaction execution remain tied to the developer account, and that users are responsible for complying with the relevant terms and legal notices. In practice, that means developers can set the rules around supported chains, approved token pairs, and other limits before an agent is allowed to act. The company also emphasizes that the user still signs transactions, which keeps the setup non-custodial rather than fully autonomous.
This approach fits 1inch’s wider strategy of making its infrastructure easier for developers to plug into existing AI workflows. Its documentation says the MCP Server works with popular tools and assistant environments, while the developer portal highlights machine-readable documentation and AI-friendly resources for builders. That puts 1inch in a position to serve teams building agent-driven products that need real-time market data, route discovery, and execution tooling rather than just static documentation.
“Agents, not humans, will be executing the majority of swaps by 2030,” according to Sergej Kunz, 1inch co-founder. “However, the agent economy cannot eliminate market competition. Trading outcomes are still defined by data and execution quality; poorly informed agents will underperform skilled humans. That is why choosing the infrastructure around the agent is as important as the strategy.”
For DeFi, the update signals a practical step toward a future where agents do more than recommend trades. They can research an action, prepare the execution path, and hand it off for approval within the same environment. 1inch is betting that this kind of infrastructure will matter as autonomous software becomes a larger part of Web3, especially for teams that want speed, live data, and controlled execution in one place.