The Graph Launches Cross-Chain Trust Infrastructure for AI Agents

The Graph's Subgraph MCP integration lets you connect Claude to over 15,000 open APIs covering major web3 protocols in about 5 minutes. **What you can do:** - Query real-time smart contract data in natural language - Access ENS records, liquidity pool data, and token mappings - Pull data from dozens of blockchain networks - Build AI agents with live onchain activity tracking **Key features:** - Works with Claude, Cursor, and Cline - No complex API calls needed - just ask questions naturally - Access to The Graph's extensive library of indexed Subgraphs The setup process is straightforward and documented in the official guide. [Get started with the setup guide](https://thegraph.com/docs/en/subgraphs/tooling/subgraph-mcp/claude/)

**Substreams Skills** now allows developers to build production-ready blockchain data pipelines by simply prompting an AI assistant. **Key capabilities:** - Deploy live, low-latency data streams from Ethereum, Base, BSC, or Robinhood contracts - Complete setup possible within a few hours - AI generates production-ready code across multiple environments **Updated framework covers:** - Development workflows - EVM and Solana chains - SQL integration - Data sinks and deployment - Testing and Portal API - Package reuse recommendations The tool emphasizes leveraging existing packages rather than building from scratch, streamlining the development process for blockchain data infrastructure. [View on GitHub](http://github.com/streamingfast/substreams-skills)

StreamingFast has released **Substreams Hosted Sinks**, a service that eliminates the need for developers to manage blockchain data infrastructure. The tool works in three steps: - Bring a Substreams package - Select a database - Click deploy StreamingFast handles the pipeline execution and delivers queryable data directly to your database. The service includes: - Low EVM latencies - Fast sync times - Automatic reorg handling - Full data ownership Developers can get started at [thegraph.market](https://thegraph.market). The team is seeking community input on which data streams to prioritize next.

Converting Subgraphs to Substreams no longer requires manual Rust coding. The process is now automated through twelve Substreams Skills available for AI coding assistants. **Key capabilities include:** - Automated Subgraph-to-Substreams conversion - Chain-specific decoding for multiple blockchains - SQL sink configuration and deployment - Built-in testing frameworks The open-source toolset integrates with popular development environments including Claude Code, Cursor, VS Code, and Windsurf. [View on GitHub](http://github.com/streamingfast/substreams-skills)

The Graph has identified four distinct trust models used in blockchain data extraction, but only one truly delivers on verifiable claims. The four models are: - **Baseline pipelines** - Standard data processing - **1-of-N** - Single source verification - **Consensus** - Agreement-based validation - **Verifiable extraction** - Cryptographically rigorous proof Only verifiable extraction meets the strict definition of "verifiable" in a technical sense. The distinction matters because providers often use the term loosely in marketing materials. When evaluating blockchain data services, understanding which trust model a provider actually implements is critical for assessing data integrity.