stream-chain

Enables multi-agent workflows with sequential data flow and context passing between steps.

• Executes multi-step agent chains with context passing between steps
• Supports custom prompt chains and predefined pipelines
• Facilitates data transformation and sequential processing
• Streams output from one agent to the next in a pipeline

• Each step receives full context from previous step output
• Custom chains allow full control over prompt sequences
• Predefined pipelines offer battle-tested workflows for common tasks
• Supports verbose, debug, and timeout controls per step
• Context flows sequentially enabling complex data transformations
• Outputs include progress, results, timing, and summary

1. 1.Use `run` to create custom chains with specific prompts
2. 2.Use `pipeline` to execute predefined workflows like analysis, refactor, test, or optimize
3. 3.Pass context between steps via sequential prompt execution
4. 4.Apply options like --verbose, --timeout, or --debug as needed
5. 5.Store reusable pipelines in config.json for reuse