faugustdev

Git Context Controller — AI skill for Claude Code

AI community

Structured context management framework for LLM agents.

How to install Git Context Controller

This entry records only its repository, not the path inside it, so there is no exact command to give. Open faugustdev/git-context-controller and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Git Context Controller does

Structured context management framework for LLM agents. Implements Git-like operations (COMMIT, BRANCH, MERGE) to manage long-horizon agent memory.

Alternatives in AI

Ccstatusline — by sirmalloc - A highly customizable status line formatter for Claude Code CLI that displays model info, git b 5.5k ★
LongHorizon Harness — The long-horizon computer-use harness 1.3k ★
Nocturne Memory — A lightweight, rollbackable, and visual Long-Term Memory Server for MCP Agents 1.3k ★

README

Git Context Controller (GCC) v2

[![Release](https://img.shields.io/github/v/release/faugustdev/git-context-controller)](https://github.com/faugustdev/git-context-controller/releases) [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE) [![Skills.sh](https://img.shields.io/badge/skills.sh-compatible-blue)](https://skills.sh) [![Available on CodeGuilds](https://img.shields.io/badge/Available_on-CodeGuilds-6366f1)](https://codeguilds.dev/packages/git-context-controller) [![DOI](https://img.shields.io/badge/DOI-10.5281/zenodo.19475089-blue)](https://doi.org/10.5281/zenodo.19475089)

**Lean, git-backed context management for LLM agents.**

GCC v2 stores **hash + intent + optional decision notes** instead of verbose markdown. Full context is reconstructed on demand via `git show`. Dual mode: git-backed or standalone.

Based on the research paper: [Git Context Controller](https://arxiv.org/abs/2508.00031)

Why GCC?

LLM agents lose context as conversations grow. GCC solves this by giving agents structured memory:

Lean storage -- ~50 tokens per entry vs ~500 in v1
Git-backed truth -- real commits, not narrative copies
Safe experimentation -- branches via git worktrees for real isolation
Cross-session recovery -- reconstruct context from hashes on demand
aiyoucli bridge -- auto-feeds commit data to vector memory when available

How It Works

Git Mode (with repo)

.GCC/
├── index.yaml       # Single source of truth (~50 tokens/entry)
├── branches/        # Branch-specific notes
├── worktrees/       # Worktree tracking
└── .bridge-log      # Sync state with aiyoucli

index.yaml entry:
  - id: C001
    hash: 85c8539        ← pointer to git truth
    intent: "release prep" ← why
    note: "descartamos semantic-release por overhead" ← optional decision
    branch: main
    date: "2026-02-25T21:40:00Z"