# PMB AI

Local memory for your coding agent: decisions, lessons and project structure live in SQLite and come back in later sessions. No cloud, no API keys.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 286
- Author: oleksiijko
- License: Apache-2.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/oleksiijko/pmb
- Updated: 2026-09-23
- HTML: https://affpapa.org/en/nevibecode/mcp/pmb-ai

## What it does

PMB AI gives an agent long-term memory. Decisions from last week, rules you taught it, facts about the project and the shape of the codebase sit in a local workspace and survive restarts, model upgrades and switching agents. SQLite is the durable source of truth, with rebuildable search indexes next to it. Nothing calls an LLM on the read path.

It installs as an MCP server. Claude and Codex take a single command; Cursor and GitHub Copilot get a JSON block pointing at `uvx pmb-ai`. Then `pmb setup` detects your agent and writes the MCP entry, `pmb warmup` preloads the model so the first recall is instant, and `pmb doctor` confirms the wiring. Hooks in Claude and Codex inject the right memory before the model starts thinking, so you never have to call a tool by hand. Search is multilingual: a Russian query finds an English fact.

To see what was captured, run `pmb dashboard` — a local web UI on `http://127.0.0.1:8765` with tabs for the entity map, timeline and lessons. It binds to `127.0.0.1` only, so nothing leaves the machine. `pmb export` dumps everything to Markdown or JSON. Apache 2.0.

## Requirements
- Needs: Python and uv (commands run via uvx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add pmb-ai -- uvx pmb-ai
```
uvx runs the command named after the package; if the README names another one, use uvx --from pmb-ai <command>

### Codex

**Platform CLI**
```bash
codex mcp add pmb-ai -- uvx pmb-ai
```
uvx runs the command named after the package; if the README names another one, use uvx --from pmb-ai <command>

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "pmb-ai": {
            "command": "uvx",
            "args": [
                "pmb-ai"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from pmb-ai <command>

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "pmb-ai": {
            "type": "stdio",
            "command": "uvx",
            "args": [
                "pmb-ai"
            ]
        }
    }
}
```
uvx runs the command named after the package; if the README names another one, use uvx --from pmb-ai <command>

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add pmb-ai uvx pmb-ai
```
uvx runs the command named after the package; if the README names another one, use uvx --from pmb-ai <command>

## When to use it
- recall an auth decision from an earlier session
- teach the agent a rule it must not break
- locate the file and line behind a bug description
- carry context over when switching agents or models
- export all project memory to Markdown

## Example prompts
- Recall what we decided about auth last week
- Find where the discount threshold is set and show me the line
- Remember: we never drop the threshold below 25%

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
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