.ai/ folder.Stop maintaining separate config files for every AI coding tool. dotai manages Claude, Gemini, Copilot & Antigravity from a single directory — plus an 8-phase Spec-Driven Development workflow that keeps AI code traceable to human-reviewed specs.
dotai creates a .ai/ directory as your source of truth, then symlinks
each file to where every AI tool expects it.
Every major AI coding tool is supported out of the box.
AI.md → CLAUDE.md, settings, commands, skills, knowledge
AI.md → GEMINI.md, settings, commands (.toml), skills, ignore, knowledge
AI.md → copilot-instructions.md, prompts, instructions, skills, knowledge
AI.md → GEMINI.md + AGENTS.md, settings, rules, workflows, skills, knowledge
AI agents reset their memory every conversation. dotai changes that — with a persistent knowledge layer that grows smarter with every commit and every session.
The result? Your AI starts every session already knowing your architecture, patterns, gotchas, and constraints.
Agents automatically persist discoveries as they work. Patterns, gotchas, and architecture decisions accumulate across sessions — no manual effort.
Every agent session starts with full codebase context from line one. No warm-up, no re-explaining. Just instant, deep understanding.
Run /learn to let your AI agent deeply analyze the codebase and populate the knowledge base. The more you use it, the smarter it gets.
The SDD Toolkit introduces an 8-phase workflow where AI agents implement features from human-reviewed specifications. No more uncontrolled AI drift.
The result? Human-quality control over AI-speed development. Review at checkpoints, not after the damage is done.
Every line of code traces back to a human-reviewed specification. No hallucinated features, no unauthorized changes.
If an evaluation fails, the agent automatically retries implementation using the diagnosis. No manual intervention needed.
28 command files generated in all 4 formats — Claude commands, Gemini commands, Copilot prompts, and Antigravity workflows.
Standard filesystem symlinks. Works out of the box.
Standard filesystem symlinks. Works out of the box.
NTFS junctions for dirs, symlinks for files. No admin needed on Win 10+.
All symlinks use relative paths for portability — move or clone the repo anywhere.
Set up once, then let the knowledge grow automatically with every commit and every AI session.
Creates .ai/ folder and symlinks for all your AI tools.
Scaffolds knowledge directory, scan skill, and /learn command.
AI agent uses MCP tools to deeply analyze and persist findings.
The more you use AI, the smarter it gets.
Common questions about using dotai in your projects.
dotai is designed for exactly this scenario. Run dotai init — it automatically detects your existing context files and merges their content into .ai/AI.md. Nothing is lost.
For existing commands and skills, copy them into .ai/, then run dotai link --force --backup. Your originals are saved as .dotai.bak files.
dotai unlink to remove all symlinks and restore originals.
No. By default, dotai skips any existing real file at a symlink target. It only replaces files when you explicitly use --force, which also creates a backup automatically.
They run dotai init. It reads .dotai.json and auto-sets up everything — no prompts needed. Similar to how npm install reads package.json.
Yes. Use dotai init --dir .custom-name to specify a custom directory. The name is stored in .dotai.json and used for all subsequent commands.
Yes. On Windows 10+, dotai uses NTFS junctions for directories and standard symlinks for files. No administrator privileges are required.
Run dotai unlink. It removes all symlinks and restores the original files from .dotai.bak backups. Your project goes back to its pre-dotai state.
Explore the full documentation — installation, configuration, knowledge base, CLI reference, and guides.
Read the Documentation