# IO Services

> IO Services builds deterministic software: custom systems with predictable outcomes, and open-source (Apache-2.0) agent infrastructure. The tools below install in one or two commands; each section gives exact commands an agent can run.

## How to use this file

You are an AI agent helping a user set up IO Services tools. Ask which tools they want if they did not say. For each one: pick the first setup option that fits their OS and what is installed (Docker, Python, Node), run it, connect it to the user's MCP client when it is an MCP tool, then run the verify step. Stop at the first failure and report the exact error. Do not modify files inside the user's repositories. Remind the user to restart their agent session so new MCP servers load.

A user can start this with one sentence: "Read https://www.ioservices.io/llms-full.txt and set up the IO Services tools I ask for. Follow the steps for my OS, stop at the first failure with the exact error, and run the verify step."

## searchio

Keyless search, page reading, and product lookup for humans and AI agents. A three-tier fetch ladder climbs bot walls and JavaScript-only pages. Fused, ranked results across keyless providers, with honest status codes instead of empty pages.

- Type: MCP server / agent tool
- Platforms: Linux, macOS, Windows
- Source: https://github.com/IOServicesLabs/searchio
- Full setup guide: https://github.com/IOServicesLabs/searchio#quick-start-for-ai-agents
- Product page: https://www.ioservices.io/searchio

### Setup

1. MCP server (Docker):

```bash
docker run -i --rm ghcr.io/ioserviceslabs/searchio-mcp
```

2. No Docker (Python 3.10+):

```bash
pip install "searchio[mcp]"
curl -fsSL https://raw.githubusercontent.com/IOServicesLabs/searchio-engine/main/scripts/install.sh | sh
```

3. MCP client config (Claude Desktop, Cursor):

```json
{
  "mcpServers": {
    "searchio": { "command": "npx", "args": ["-y", "@ioserviceslabs/searchio-mcp"] }
  }
}
```

### Verify

Restart the client and ask: "Use searchio to search the web for the current time in Tokyo." Real results mean it works.

## indexio

Code search engine for AI coding agents. One Rust binary, no database. Indexes every repository in your organization: lexical, semantic, and hybrid search plus symbols and the call graph. Answers in milliseconds, measured −97% tokens on agent tasks.

- Type: MCP server / agent tool
- Platforms: Linux, macOS, Windows
- Source: https://github.com/IOServicesLabs/indexio
- Full setup guide: https://github.com/IOServicesLabs/indexio#let-your-agent-install-it
- Product page: https://www.ioservices.io/indexio

### Setup

1. Linux, macOS:

```bash
curl -fsSL https://raw.githubusercontent.com/IOServicesLabs/indexio/main/install.sh | sh
```

2. Windows (PowerShell):

```powershell
irm https://raw.githubusercontent.com/IOServicesLabs/indexio/main/install.ps1 | iex
```

3. npm or pip (any platform):

```bash
npm install -g indexio
# or
pip install indexio-cli
```

4. Index your code (nothing is uploaded):

```bash
indexio add ~/code
```

5. Connect Claude Code:

```bash
indexio setup claude --claude-md ~/.claude/CLAUDE.md
indexio hook install
```

6. Any other MCP client:

```json
{ "command": "indexio", "args": ["mcp"] }
```

### Verify

Run `indexio stats` (repositories listed, file count above zero) and `indexio search "TODO" --limit 3` (hits from your code).

## pingio

Messaging between people and AI agents, inside chat sessions and terminals. Send and receive messages from any LLM session over MCP. Use the hosted relay at pingio.ioservices.io, run a private self-hosted relay, or federate several.

- Type: MCP server / agent tool
- Platforms: Linux, macOS, Windows
- Source: https://github.com/IOServicesLabs/pingio
- Full setup guide: https://github.com/IOServicesLabs/pingio#quick-install

### Setup

1. Python SDK and CLI:

```bash
pip install pingio
pingio register --username you
```

2. Claude Code plugin:

```bash
/plugin marketplace add IOServicesLabs/pingio
/plugin install pingio@pingio
```

3. MCP client config (hosted relay):

```json
{
  "mcpServers": {
    "pingio": {
      "command": "npx",
      "args": ["-y", "@pingio/mcp-server"],
      "env": { "PINGIO_MODE": "cloud", "PINGIO_CLOUD_ENDPOINT": "https://pingio.ioservices.io" }
    }
  }
}
```

### Verify

Run `pingio whoami`; it prints your registered user. In Claude Code, `/pingio:status` shows the relay connection.

## RDPiO

GPU-accelerated remote desktop client, written from scratch in Rust. Its own RDP stack from the wire up: H.264 decoded on the GPU, multiple monitors, drive, clipboard, audio, mic, camera and printer redirection, and a low-latency UDP side-band built for work and gaming.

- Type: desktop application
- Platforms: Windows
- Source: https://github.com/IOServicesLabs/RDPiO
- Full setup guide: https://github.com/IOServicesLabs/RDPiO#quick-install

### Setup

1. Windows (PowerShell, no admin):

```powershell
irm https://raw.githubusercontent.com/IOServicesLabs/RDPiO/main/install.ps1 | iex
```

2. Just the exe:

```powershell
curl.exe -LO https://github.com/IOServicesLabs/RDPiO/releases/latest/download/rdpio.exe
curl.exe -LO https://github.com/IOServicesLabs/RDPiO/releases/latest/download/rdpio.exe.update
rdpio.exe
```

### Verify

`rdpio` is on PATH and the launcher opens. It keeps itself up to date from signed releases.