# AitherOS ADK

Connects an agent runtime with SQLite knowledge graphs, effort-based routing, and subagent delegation. Useful for keeping long-term memory across chats and dispatching tasks to helper agents.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 11
- Author: Aitherium
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/Aitherium/awdk
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/mcp/aitheros-adk

## What it does

The runtime serves as an MCP bridge executed via `uvx aither-adk`. It maintains an embedded knowledge store inside a local SQLite database. During conversations, entities and connections get parsed and saved automatically. Search queries combine classic keyword matching with semantic retrieval. No external graph databases are required.

Model dispatch relies on an internal effort rating scaled from 1 to 10. Routine questions go to lightweight engines, while heavy analytical tasks reach deep reasoning systems. You can swap execution backends during runtime between local runtimes like llama.cpp and cloud providers such as Anthropic or OpenAI. Multiple agents coordinate using the built-in `ask_agent` capability.

Without valid `AITHERIUM_API_KEY` and `AITHERIDENTITY_URL` variables, the server fails to authenticate. On Windows machines, Python script directories might also need manual registration inside your user `PATH`.

## Requirements
- Needs: Python and uv (commands run via uvx)
- Environment variables: `AITHERIDENTITY_URL`, `AITHERIUM_API_KEY` — keys and tokens: ask the human for the values, never commit them or put them in shared files.

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add --env AITHERIDENTITY_URL=<AITHERIDENTITY_URL> --env AITHERIUM_API_KEY=<AITHERIUM_API_KEY> --transport stdio aitheros-adk -- uvx aither-adk
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from aither-adk <command>

### Codex

**Platform CLI**
```bash
codex mcp add aitheros-adk --env AITHERIDENTITY_URL=<AITHERIDENTITY_URL> --env AITHERIUM_API_KEY=<AITHERIUM_API_KEY> -- uvx aither-adk
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from aither-adk <command>

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "aitheros-adk": {
            "command": "uvx",
            "args": [
                "aither-adk"
            ],
            "env": {
                "AITHERIDENTITY_URL": "<AITHERIDENTITY_URL>",
                "AITHERIUM_API_KEY": "<AITHERIUM_API_KEY>"
            }
        }
    }
}
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from aither-adk <command>

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "aitheros-adk": {
            "type": "stdio",
            "command": "uvx",
            "args": [
                "aither-adk"
            ],
            "env": {
                "AITHERIDENTITY_URL": "<AITHERIDENTITY_URL>",
                "AITHERIUM_API_KEY": "<AITHERIUM_API_KEY>"
            }
        }
    }
}
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from aither-adk <command>

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add -e AITHERIDENTITY_URL=<AITHERIDENTITY_URL> -e AITHERIUM_API_KEY=<AITHERIUM_API_KEY> aitheros-adk uvx aither-adk
```
Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from aither-adk <command>

## When to use it
- Store conversational entities in a local SQLite knowledge graph
- Route prompts across models according to task difficulty scores
- Delegate subtasks to specialized agents using the `ask_agent` tool
- Switch inference backends on the fly between local engines and cloud APIs
- Query historical context with combined semantic and keyword search

## Example prompts
- Search the knowledge graph for previous notes on our architecture
- Hand this debugging task over to a subagent using `ask_agent`
- Evaluate task complexity and assign a high-effort model for code refactoring

## Similar
- [alaya-mcp](https://affpapa.org/en/nevibecode/mcp/alaya-mcp.md) — MCP server · ★ 18
- [Local Memory](https://affpapa.org/en/nevibecode/mcp/local-memory.md) — MCP server · ★ 15
- [kairn](https://affpapa.org/en/nevibecode/mcp/kairn.md) — MCP server · ★ 13
- [Waggle-mcp](https://affpapa.org/en/nevibecode/mcp/waggle-mcp.md) — MCP server · ★ 44
- [yantrikdb-mcp](https://affpapa.org/en/nevibecode/mcp/yantrikdb-mcp.md) — MCP server · ★ 25
- [agentdb-query](https://affpapa.org/en/nevibecode/skills/agentdb-query.md) — skill · 745 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md