git-lfs skill

Overview

This skill integrates Git LFS workflows into agent-driven game development pipelines to manage large binary assets reliably. I built it to handle setup, tracking, storage operations, and migration tasks so agents can treat big files like first-class repo objects. It works alongside babysitter orchestration to enable deterministic, resumable operations for asset-heavy projects.

How this skill works

The skill automates LFS repository initialization, configures tracked patterns, and creates or updates .gitattributes entries so agents can commit large files without bloating Git history. It exposes commands for querying LFS storage usage, pruning objects, and performing migrations from legacy storage to LFS. All operations are designed to be idempotent and resumable so orchestration can recover from interruptions.

When to use it

• Your repo contains large binary game assets (textures, audio, models) that exceed normal Git limits.
• You need agents to commit, track, and manage large files deterministically within automated pipelines.
• You want to migrate an existing repository’s large objects into Git LFS without losing history.
• You must monitor and control LFS storage quotas and prune unused objects safely.

Best practices

• Define clear .gitattributes patterns early (by file extension or folder) to avoid accidentally committing binaries to Git.
• Run LFS migrations in a staging environment and validate artifacts before rewriting main branches.
• Use resumable, idempotent commands so automation can retry without causing duplicate work.
• Schedule periodic pruning and storage audits to keep repository sizes manageable.
• Keep credentials and remote LFS endpoints secured and rotated as part of CI secrets management.

Example use cases

• Automatically add and track new art asset types committed by content pipelines during build runs.
• Migrate a legacy game repository’s historical large files into LFS while preserving commit history.
• Enforce storage limits by detecting heavy LFS usage and triggering cleanup or alerts.
• Resume interrupted LFS upload sequences in a multi-step agent orchestration without data loss.

FAQ