# Prompt: Install Aleph Across All Environments
Use this prompt with Claude Code (or any AI assistant with file access) to install Aleph on all your AI coding tools.
---
## Prompt
```
I need you to install and configure Aleph (an MCP server for recursive LLM reasoning over large local data) across all my AI development environments.
**Target environments:**
1. Claude Code (CLI) - via ~/.claude/claude_desktop_config.json (macOS/Linux) or %USERPROFILE%\.claude\claude_desktop_config.json (Windows)
2. Claude Desktop - via ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)
3. Cursor - via cursor MCP config
4. Windsurf - via windsurf MCP config
5. Codex CLI - via ~/.codex/config.toml (macOS/Linux) or %USERPROFILE%\.codex\config.toml (Windows)
6. Gemini CLI - via ~/.gemini/mcp.json
7. Kimi CLI - via ~/.kimi/mcp.json
**Tasks:**
1. First, check if aleph-rlm is installed:
```bash
pip show aleph-rlm || pip install "aleph-rlm[mcp]"
```
2. Run the automatic installer and see what it detects:
```bash
aleph-rlm install
```
3. For any environments not auto-detected, manually configure:
**Claude Desktop:**
- macOS: `~/Library/Application Support/Claude/claude_desktop_config.json`
- Windows: `%APPDATA%\Claude\claude_desktop_config.json`
```json
{
"mcpServers": {
"aleph": {
"command": "aleph",
"args": ["--enable-actions", "--tool-docs", "concise", "--workspace-mode", "any"]
}
}
}
```
**Codex CLI:**
- macOS/Linux: `~/.codex/config.toml`
- Windows: `%USERPROFILE%\.codex\config.toml`
```toml
[mcp_servers.aleph]
command = "aleph"
args = ["--enable-actions", "--tool-docs", "concise", "--workspace-mode", "any"]
```
**Gemini CLI** (~/.gemini/mcp.json):
```json
{
"mcpServers": {
"aleph": {
"command": "aleph",
"args": ["--enable-actions", "--tool-docs", "concise", "--workspace-mode", "any"]
}
}
}
```
**Kimi CLI** (~/.kimi/mcp.json):
```json
{
"mcpServers": {
"aleph": {
"transport": "stdio",
"command": "aleph",
"args": ["--enable-actions", "--tool-docs", "concise", "--workspace-mode", "any"]
}
}
}
```
4. Install skills (optional but recommended):
Download `docs/prompts/aleph.md` from the repo and save it to:
**Claude Code:**
- macOS/Linux: `~/.claude/commands/aleph.md`
- Windows: `%USERPROFILE%\.claude\commands\aleph.md`
**Codex CLI:**
- macOS/Linux: `~/.codex/skills/aleph/SKILL.md`
- Windows: `%USERPROFILE%\.codex\skills\aleph\SKILL.md`
**Kimi CLI (recommended):**
- User-level: `~/.config/agents/skills/aleph/SKILL.md`
- Project-level: `.agents/skills/aleph/SKILL.md`
Or from the installed package:
```bash
# macOS/Linux
mkdir -p ~/.codex/skills/aleph
cp "$(python -c "import aleph; print(aleph.__path__[0])")/../docs/prompts/aleph.md" ~/.codex/skills/aleph/SKILL.md
# Kimi CLI (user-level)
mkdir -p ~/.config/agents/skills/aleph
cp "$(python -c "import aleph; print(aleph.__path__[0])")/../docs/prompts/aleph.md" ~/.config/agents/skills/aleph/SKILL.md
```
5. Configure `sub_query` (optional):
If you want the API backend, set environment variables:
**macOS/Linux** (add to ~/.zshrc or ~/.bashrc):
```bash
export ALEPH_SUB_QUERY_API_KEY=sk-...
export ALEPH_SUB_QUERY_MODEL=your-model-name
# Optional: OpenAI-compatible base URL (Groq, Together, local LLMs, etc.)
export ALEPH_SUB_QUERY_URL=https://api.your-provider.com/v1
```
**Windows** (set via System Environment Variables or PowerShell $PROFILE):
```powershell
$env:ALEPH_SUB_QUERY_API_KEY = "sk-..."
$env:ALEPH_SUB_QUERY_MODEL = "your-model-name"
```
CLI backends (`claude`, `codex`, `gemini`) do not require an API key. Aleph will auto-detect the first available backend unless you set `ALEPH_SUB_QUERY_BACKEND`.
6. Verify installation:
```bash
aleph-rlm doctor
```
7. Test by restarting each environment and checking if the aleph tools are available.
**Important notes:**
- The `--enable-actions` flag allows file read/write and command execution
- `--enable-actions` also enables rg-based search, smart loaders, and memory pack auto-save
- In CLI environments (Claude Code, Codex, Gemini), `sub_query` can use the local CLI backend - no API key needed
- The installer should handle most of this automatically, but verify each environment works
- For per-project scoping, set `--workspace-root /absolute/path/to/project` instead of `--workspace-mode git`. See MCP_SETUP.md for details.
- Some MCP clients don't reliably pass `env` vars from config to the server process. If `sub_query` reports missing credentials, add exports to your shell profile and restart the client.
Please proceed step by step and report what you find in each environment.
```
---
## What this does
- Installs `aleph-rlm` Python package
- Configures MCP server in all supported AI coding environments
- Optionally sets up OpenAI-compatible API credentials for `sub_query`
- Enables action tools (file I/O, commands) where appropriate
