# Content Interviewer MCP

Content Interviewer MCP is a local Python MCP server that discovers research topics, interviews you to uncover your unique perspective, and turns your insights into content such as short-form Reel scripts.

## Project status

The MVP is complete and ready for local use with MCP-compatible clients such as Codex.

## Requirements

- Python 3.11 or newer
- SQLite with FTS5 support

## Development setup

Create a virtual environment:

```bash
python3 -m venv .venv
```

Activate it:

```bash
source .venv/bin/activate
```

Install the project and its development dependencies:

```bash
python -m pip install -e ".[dev]"
```

## Development checks

Run the tests:

```bash
python -m pytest
```

Run linting:

```bash
python -m ruff check .
```

Check that the source compiles:

```bash
python -m compileall -q src
```

## Architecture

The initial version uses:

- the official Python MCP SDK
- Pydantic for input validation
- SQLite for local persistence
- SQLite FTS5 for full-text search
- pytest for automated testing
- Ruff for linting
- STDIO for communication with Codex

The project starts as a local, single-user application. SQLite keeps private data local without requiring a separate database server. FTS5 provides the initial search implementation; embeddings will only be considered after the MVP demonstrates a need for them.

## Project structure

```text
src/
  content_interviewer/
    server.py
    config.py
    db/
      connection.py
      migrations/
      repositories/
    services/
    tools/
    schemas/
    utils/
    resources/
tests/
fixtures/
docs/
data/
```

## Privacy

ChatGPT exports, SQLite databases, environment files, logs, and other private local data must not be committed to version control. The server must not log complete imported conversations or article bodies.

## MVP boundaries

Version 0.1 does not include a web dashboard, multi-user accounts, remote deployment, embeddings, automatic social-media publishing, or live access to a ChatGPT account.