# Consensus Computing in CRE Source: https://docs.chain.link/cre/concepts/consensus-computing Last Updated: 2025-11-04 > For the complete documentation index, see [llms.txt](/llms.txt). **Consensus computing** is the foundational computing paradigm that makes CRE secure and reliable. It ensures that every operation your workflow performs—whether fetching data from an API or reading from a blockchain—is verified by multiple independent nodes before producing a final result. ## What is consensus computing? Consensus computing is when a decentralized network of nodes must form consensus as part of executing code and storing information. Unlike traditional computing where you trust a single server or service, consensus computing provides unique guarantees: - **Tamper-resistance**: No single node can manipulate results - **High availability**: The network continues operating even if individual nodes fail - **Trust minimization**: You don't need to trust any single entity - **Verifiability**: All results are cryptographically verified Blockchains pioneered consensus computing for maintaining asset ledgers and executing smart contracts. CRE extends this paradigm to **any offchain operation**—API calls, computations, and more. ## How CRE uses consensus In CRE, **every execution capability automatically includes decentralized consensus**. Here's how it works: 1. **Independent execution**: When your workflow invokes a capability (like `http.Client` or `evm.Client`), each node in the dedicated capability DON performs the operation independently 2. **Result collection**: Each node produces its own result based on what it observed 3. **Consensus protocol**: The DON applies a Byzantine Fault Tolerant (BFT) consensus protocol to validate and aggregate the individual results 4. **Verified output**: A single, consensus-verified result is returned to your workflow This process happens automatically for every capability call. You don't need to write any special code—consensus is built into the CRE runtime environment. ## Why this matters for your workflows ### Protection against node failures and manipulation CRE's consensus model protects against individual node failures and malicious behavior. When multiple independent nodes execute the same operation: - **Node-level resilience**: If some nodes fail or go offline, the network continues operating - **Byzantine Fault Tolerance**: Even if some nodes are compromised and return incorrect results, the honest majority ensures the correct outcome - **Execution consistency**: All nodes must execute your workflow logic identically, preventing manipulation by individual operators ### Validated and verified results Every result your workflow receives has been cryptographically verified and validated across multiple nodes. This provides strong guarantees that: - The operation was executed correctly - The result matches what independent observers agreed upon - No single node operator can manipulate your workflow's execution or outputs ### Unified security model With CRE, your **entire institutional-grade smart contract**—not just the onchain parts—benefits from consensus computing. This means: - **API responses** are validated across multiple nodes before your workflow uses them - **Blockchain reads** are verified by multiple nodes - **Blockchain writes** are validated by multiple nodes before being submitted onchain - **Computation results** within your workflow are executed consistently across all nodes Your workflow inherits the same security and reliability guarantees as blockchain transactions, but for any offchain operation. ## Consensus in practice ### HTTP capability When your workflow makes an API request using `http.Client`, the HTTP capability DON executes your request across multiple nodes. Their responses are validated through a consensus protocol before returning a result to your workflow. This ensures: - Execution consistency across all nodes - Protection against individual node compromise or failure - Detection of inconsistent API responses (e.g., due to load balancing or timing) See the [HTTP Capability](/cre/capabilities/http) page for details. ### EVM Read & Write capability When your workflow reads from or writes to a blockchain using `evm.Client`, the EVM capability DON performs the operation across multiple nodes: - **For reads**: Multiple nodes independently query the blockchain, and consensus validates their responses match - **For writes**: Multiple nodes agree on the transaction data before submitting it onchain This ensures: - Execution consistency across all nodes - Protection against individual node compromise or failure - Validated blockchain data before use in your workflow See the [EVM Read & Write Capability](/cre/capabilities/evm-read-write) page for details. ## Consensus in simulation vs. deployed workflows > **NOTE: Understanding consensus during development** > > **In local simulation** (`cre workflow simulate`), consensus uses a single-node model to test your workflow's logic quickly. The "consensus" step still occurs, but involves a single node wrapping its result in a standardized report structure. This allows you to build and test against the final SDK interfaces without the overhead of multi-node consensus. **In deployed workflows**, true multi-node Byzantine Fault Tolerant (BFT) consensus is performed across all capability DON operations, providing full decentralized security guarantees. ## Learn more Learn more about how to use CRE capabilities with built-in consensus: - **[Capabilities Overview](/cre/capabilities)**: Explore all available capabilities - **[API Interactions](/cre/guides/workflow/using-http-client)**: Learn how to use the HTTP capability with built-in consensus - **[EVM Chain Interactions](/cre/guides/workflow/using-evm-client/overview)**: Learn how to use the EVM capability with built-in consensus