Alpha 1
Run your build and test workflows locally.
Loom is an open-source workflow engine for local host and Docker jobs. Define your commands in YAML, run them with Loom, and use its runtime artifacts to inspect failed jobs and steps.
Inspect a failed command
When a job fails, read its state and match its job, step and command IDs in event messages. Read that output to investigate the cause. If recording could not start, use the terminal error.
Working with run results
Read state.jsonl to identify a failed operation, then match its IDs in
events.jsonl. Use loom logs or MCP read tools.
Share the run ID, relevant state records and a short reviewed event excerpt with a teammate.
MCP reads the same state and event files. Event filters select a job or stream; file size depends on the workflow and the output it produces.
The workflow
Put your build and test commands in .loom/workflow.yml. Run loom check to validate its structure, then loom run --local to execute it. Jobs can use the host or a Docker image. The Hello Loom guide walks through a complete example.
Alpha 1 scope
Alpha 1 includes local execution with host and Docker providers on Linux, caching, secrets, template composition, and structured runtime artifacts. Artifact contents vary between runs, and contracts may change between releases. Remote execution is planned. Read the support boundary before building on Alpha behavior.
Explore
Core concepts, artifact model, and Alpha 1 scope.
Use cases and limits for local workflows and failure diagnosis.
Workflow execution and the files used to investigate failures.
Set up Loom MCP, the Cursor skill, and agent-driven workflows.
Run the Hello Loom example
Create a small workflow, run it, then inspect the output from an intentional failure.
Start with Hello Loom