Getting Started

Get Logsmith running in under 5 minutes.

Prerequisites

-Claude Code (free CLI) or Claude Desktop — get it from claude.ai/download
-A Logsmith API key from your dashboard
-ffmpeg installed on your machine (for frame extraction)

1. Install the Plugin

For Claude Code, run:

claude plugin add logsmith

This installs the Logsmith MCP server and labeling skills. The plugin auto-connects to our intelligence server for alignment, vision, and labeling.

2. Set Your API Key

Copy your API key from the dashboard and set it as an environment variable:

export LOGSMITH_API_KEY=lsk_live_your_key_here

Add this to your shell profile (~/.zshrc or ~/.bashrc) so it persists across sessions.

3. Label Your First Footage

Open Claude Code in the directory with your video clips:

cd /path/to/your/clipsclaude

Then tell Claude what you're working with:

"Label this footage. It's BYU football vs Arizona from October 12."

Logsmith handles the rest: it extracts metadata, fetches play-by-play data, aligns clips to plays, analyzes frames with AI vision, and generates labels for your approval. You review, correct if needed, and approve. Files get renamed with rich metadata filenames.

What Logsmith Produces

-Renamed files with structured metadata filenames (sport, opponent, period, players, play type, quality)
-JSON sidecars (.logsmith.json) with full label data, confidence scores, vision descriptions, and matched play info
-Checkpoint files for resumable sessions — pick up where you left off if a session is interrupted

Supported Sports

Football

Full play-by-play matching, down/distance, scoring plays, player roles

Basketball

Shot types, positions, fast break detection, period-based alignment

Golf

Tournament footage, tee box identification, player tracking

Event

General event footage with no play-by-play. Vision-based labeling with freeform descriptions.

More sports coming. Enterprise customers can request custom profiles.