stream-chain

Stream-Chain coordinates multi-agent pipelines by streaming outputs from one step to the next, enabling each agent's output to become context for the following step. Use `run` for custom chains or `pipeline` for predefined workflows across code analysis, security audits, testing, and refactoring tasks.

Custom chains let you define prompt sequences with full control via `run`, while predefined pipelines provide battle-tested workflows via `pipeline`. Both propagate full context history through each step, but custom chains offer flexibility and pipelines offer proven patterns for common workflows.

Yes, Stream-Chain applies to code analysis, security audits, testing, and refactoring by chaining multi-agent steps where each agent processes the previous step's output. This enables complex data transformations with deterministic, sequential processing across your security and analysis workflows.

Stream-Chain provides comprehensive execution metadata including progress tracking, timing information, and results for each step. Configure verbose mode, timeout settings, and debug options to monitor and control your pipeline execution in real time.

Stream-Chain requires prompt-based step definitions and supports configurable options like verbose logging, timeout limits, and debug mode. No external dependencies are needed; you define your workflow steps and the Skill handles streaming coordination and metadata collection.