specdriver.ai
AI IDE Config & Specs via MCP

One platform, every AI editor

Manage settings, sync configurations, and author specs that translate to every AI IDE's native format — all via MCP.

How It Works

Get started in 4 steps

Connect any AI IDE to SpecDriver via MCP and start managing config files and specs.

1
Sign in & connect your AI IDE
Sign in with GitHub, copy the MCP config for your editor, and paste it into your IDE's MCP settings.
2
Run sd_init in your project
Register your project with SpecDriver. Creates sd_workspace.cfg with your workspace info.
3
Sync settings with sd_config_pull / sd_config_push
Pull or push editor-specific config files. Each AI tool gets its own native format.
4
Manage specs with sd_spec_pull / sd_spec_push
Author one global spec, pull it translated to your editor's native format. Backs up local files automatically.
Editor Support

Native formats for every AI IDE

SpecDriver translates your spec into each editor's native format automatically.

Claude Code
CLAUDE.md
settings.json
mcp.json
Kiro
steering.md
settings.json
mcp.json
Codex
AGENTS.md
settings.json
Copilot
copilot-instructions.md
settings.json
Gemini CLI
GEMINI.md
settings.json

Welcome, User

Your AI IDE configuration hub

Tier Basic
Build v0.4.1774251680
MCP 0 connected

MCP Connections

Loading connections...

Recent Activity

Loading activity...

Admin Dashboard

Admin
Theme:

Users

UserTierSpecsConfigsLast LoginAdmin

Feature Management

Features are determined by user tier. Change a user's tier above to adjust their access.

FeatureFreeDeveloperNo Billing
MCP Access
Custom Skills
Config File Sync
Spec Create / Edit
GitHub Sync
Private Specs
Protected Skills

Recent Activity

⚡ MCP Connection

Select your editor, then copy the config into your IDE.

Endpoint
https://mcp.specdriver.ai
MCP Config (paste into mcp.json)
{}
MCP Tools

Config

SpecDriver Settings

Global settings for your SpecDriver account. Enable or disable AI IDEs and configure your MCP connection.

AI IDEs

Enable the AI IDEs you use. Disabled IDEs are hidden from the dock and their config files are not synced.

Claude Code
Kiro
Codex Beta
Copilot Beta
Gemini CLI Beta
API Key Management

API keys authenticate MCP requests (sd_config_pull, sd_config_push). Create named keys for different machines or workflows.

Skills

Create and manage custom skills that get loaded into your AI IDE context via MCP. Skills are shared across all AI tools.

Skill QuickStart

Loading...

Config Files

Manage your AI IDE configuration files stored in SpecDriver. Edit each file and save to sync via MCP.

Specs

Edit Spec

Plans & Tiers

Choose the plan that fits your workflow. Start free, upgrade when you need GitHub sync and private specs.

Free
specdriver basic
$0 /month
 
  • Config file management
  • Push/pull configs via MCP
  • Custom skills
  • Spec upload & AI conversion
  • No GitHub integration
  • Public specs only

Documentation

Learn how SpecDriver works — manage AI IDE configuration files and sync them via MCP.

What is SpecDriver?

SpecDriver is a cloud service that stores, manages, and syncs configuration files for AI-powered code editors. Instead of manually maintaining config files across machines and projects, SpecDriver lets you manage them from a single dashboard and sync them via MCP.

Supported AI IDEs

  • Claude Code — CLAUDE.md, ~/.claude/settings.json, ~/.claude/mcp.json
  • Kiro — .kiro/steering/*.md, .kiro/settings/mcp.json
  • Codex — AGENTS.md (Beta)
  • Copilot — .github/copilot-instructions.md (Beta)
  • Gemini CLI — GEMINI.md (Beta)

How It Works

1. Sign in with GitHub and select your AI IDEs from the dock.

2. Configure your config files (steering, settings, MCP) in the SpecDriver dashboard.

3. Connect your AI IDE to SpecDriver via MCP using your personal endpoint.

4. Sync — Run sd_status to check sync status. Use sd_config_sync to pull configs.

Config Files by Editor

Claude Code

FilePathPurpose
CLAUDE.md./CLAUDE.mdProject context and steering loaded every session
settings.json~/.claude/settings.jsonGlobal permissions, preferences, environment vars
mcp.json~/.claude/mcp.jsonMCP server definitions
Project settings.claude/settings.jsonPer-project overrides (committed to repo)

Kiro

FilePathPurpose
Steering.kiro/steering/*.mdSteering files with YAML frontmatter
MCP Config.kiro/settings/mcp.jsonMCP server definitions for Kiro

Codex (Beta)

FilePathPurpose
AGENTS.md./AGENTS.mdAgent instructions for Codex

Copilot (Beta)

FilePathPurpose
Instructions.github/copilot-instructions.mdRepository-level Copilot instructions

Skills

Skills are custom guidelines that get loaded into your AI IDE context via MCP. Create skills to define project conventions, coding standards, or any instructions you want your AI tools to follow. Skills are shared across all AI tools.

MCP Tools Reference

SpecDriver exposes the following tools via MCP (Model Context Protocol):

ToolDescription
sd_helpList all available SpecDriver tools with descriptions and usage
sd_statusCheck auth status, user info, and config file sync status
sd_config_pullPull all synced config files for the current editor
sd_config_pushPush a config file to SpecDriver (config_type + content)

Connection URL

Your MCP endpoint follows this format:

https://mcp.specdriver.ai/{username}/{editor}

Examples:

  • https://mcp.specdriver.ai/tonynv/claude
  • https://mcp.specdriver.ai/tonynv/kiro

Quick Start

1. Add SpecDriver to your MCP config with your API key (see Quick Connect page).

2. Run sd_help to see all available tools.

3. Run sd_status with your GitHub username to check sync status.

4. Run sd_config_pull to download your config files.

5. Run sd_config_push to upload local config files to SpecDriver.

Plans & Access Tiers

SpecDriver Basic (Free)

Config file management for all supported AI IDEs. MCP access, custom skills, and config sync. Perfect for getting started.

SpecDriver Developer ($5/month)

GitHub sync, private specs, protected skills, and advanced config management across machines.

Admin Billing

Platform administrators can set any user to nobilling or assign a custom rate.