Browse / General-purpose / ai-workspace

ai-workspace

dmytroivasikiv

A ready-to-clone control plane for Claude Code and Codex.

Open dmytroivasikiv/ai-workspace → On the boundary of the four rules.

Works with any agent

ships AGENTS.md, the cross-tool instruction file.

In its own words

README.mddmytroivasikiv/ai-workspace

A ready-to-clone control plane for Claude Code and Codex. It reduces repeated context through scoped memory retrieval, project isolation, reusable skills, and advisory lifecycle hooks without forcing a cheaper model or lower reasoning.

The repository includes two synthetic projects, six fictional colleagues, project memory, a task handoff, runnable tests, a repository-local Claude hook and skill, an Obsidian vault, QMD integration, and on-demand Graphify commands.

Quick start

Requirements: Git, Python 3.10+, Node.js 20+, and QMD.

To build the optional semantic index after QMD downloads its local model, run ./bin/aiw setup --semantic.

Open memory/ as an Obsidian vault if you want the visual knowledge view.

Enable Claude and Codex everywhere

Preview the user-level changes:

Then install the shared rules, skills, Claude output style, lifecycle hooks, and stable helper commands once per machine:

The installer preserves existing Claude and Codex rule files, merges Claude settings, and creates a settings backup. Root .claude/settings.json intentionally has no shared hooks: project hooks only load from the session's starting directory, while the installed user hook works from every registered repository.

Add a local project

Keep repositories beside this workspace or elsewhere on your machine. Their paths live in ignored workspace.yaml, never in the public template.

Quoting 48% of the README. Read the rest on GitHub ↗

Read from the repository's git tree on 2026-08-17.