Summary

viewllm addresses the challenge of viewing and sharing rich HTML/Markdown reports generated by AI coding agents, a gap left by tools like VS Code Live Preview and GitHub. It offers developers a single-binary, zero-install solution that works seamlessly across terminals, SSH, and WSL, enabling immediate access and collaboration on AI-generated artifacts without complex setup.

How It Works

This project is built around a single, statically linked Go binary that embeds the entire frontend. It efficiently scans specified directories for .html and .md files, serving a responsive web UI that polls for changes every two seconds. Markdown content is rendered client-side using marked.js, ensuring a lightweight and dependency-free operation. This approach prioritizes simplicity, minimal resource consumption (~7MB memory), and rapid startup (~100ms).

Quick Start & Requirements

The primary installation method is npx viewllm@latest, providing zero-install access to the latest version. Alternatively, binaries can be downloaded from GitHub Releases, or the project can be built from source using go build. No non-default prerequisites are required beyond Node.js for npx.

Highlighted Details

• Extremely lightweight: ~9MB binary, ~7MB memory usage, ~100ms startup time.
• Real-time file watching with per-device unread tracking and hover previews.
• Optional public URL sharing via Cloudflare Tunnel (-tunnel) without port forwarding.
• Supports HTML and Markdown rendering, themes (Light, Dark, Solarized), and mobile view.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), or roadmap were found in the provided README.

Licensing & Compatibility

The project is released under the MIT license, permitting broad use, modification, and distribution, including for commercial purposes and integration into closed-source projects.

Limitations & Caveats

File system changes are detected via client-side polling every two seconds, which may introduce a slight delay in updates. Markdown rendering occurs client-side, relying on the marked.js library. Publicly shared links generated with -tunnel are ephemeral and expire upon stopping the viewllm process.