Loom addresses the challenge of reliably delivering software built by AI coding agents, transforming them from demo creators into repeatable delivery systems. It targets developers and power users leveraging agents like Claude Code, Codex, and OpenCode, providing a structured harness to manage the end-to-end software lifecycle from idea to release, thereby reducing wasted tokens and manual effort.

How It Works

Loom functions as an open-source delivery harness, orchestrating coding agents through a structured loop: planning, building, verification, repair, preview, and handoff. It employs dynamic workflows and persists critical delivery state—including project context, task contracts, backend status, and test results—within the .loom/ directory. This stateful approach allows delivery processes to survive interruptions, agent switches, and session resets, offering a durable alternative to fragile, one-shot prompt chains and mitigating common agent failure modes like goal drift and partial completion.

Quick Start & Requirements

• Prerequisites: Node.js (>= 20), npm, Docker (for deployment previews).
• Installation: Clone the repository, run npm install, then install necessary coding agent adapters via npm run plugin:install-codex, npm run plugin:install-claude, npm run plugin:install-opencode, or npm run plugin:install-adapters.
• Verification: Use the stable CLI launcher $HOME/.loom/bin/loom-cli --version or $HOME/.loom/bin/loom-cli status.
• Usage: Initiate deliveries via agent commands (e.g., @loom build in Codex, /loom build in Claude Code) or the CLI. Use continue commands to resume interrupted workflows.
• Docs: Official quick-start and usage guides are integrated within the README.

Highlighted Details

• Dynamic workflows adapt delivery paths for each goal.
• Durable harness persists state, enabling resumption and multi-agent continuity.
• Token-saving context management reduces redundant repository reads.
• Task contracts define bounded work items with clear acceptance criteria.
• Agent-neutral CLI and multi-agent protocol support interoperability.
• Integrated verification loops include testing, logging, and repair mechanisms.
• Tracks backend readiness and UIX guidance as core delivery requirements.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or roadmap are provided in the README.

Licensing & Compatibility

• License: Apache License 2.0.
• Compatibility: The Apache 2.0 license is generally permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

Production deployment is explicitly stated as not yet supported; current deployment capabilities are limited to local Docker Compose previews, validation, logging, and repair guidance.