Metadata-Version: 2.4
Name: phileas-memory
Version: 0.1.0
Summary: Local-first long-term memory for AI companions
Project-URL: Homepage, https://github.com/alexajuno/phileas
Project-URL: Repository, https://github.com/alexajuno/phileas
Project-URL: Issues, https://github.com/alexajuno/phileas/issues
License-Expression: MIT
Keywords: ai,embeddings,llm,local-first,mcp,memory,rag
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Programming Language :: Python :: 3.14
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.11
Requires-Dist: chromadb>=1.0.0
Requires-Dist: click>=8.3.1
Requires-Dist: fastapi>=0.136.3
Requires-Dist: kuzu>=0.8.0
Requires-Dist: litellm>=1.83.0
Requires-Dist: mcp[cli]
Requires-Dist: rich>=14.3.3
Requires-Dist: sentence-transformers>=5.3.0
Requires-Dist: uvicorn>=0.42.0
Provides-Extra: embeddings
Requires-Dist: sentence-transformers; extra == 'embeddings'
Description-Content-Type: text/markdown

# Phileas — persistent memory for AI

AI conversations reset every session. Phileas is a local memory layer that any LLM can read and write through [MCP](https://modelcontextprotocol.io/), so context survives across sessions and tools.

## Get started

```bash
pip install phileas-memory
phileas init
```

The setup wizard connects Phileas to your MCP client (Claude, GPT, Ollama, or any other) and chooses where to store memories.

### First run

On first run, `phileas init` downloads two small models that run locally — an
embedding model (`all-MiniLM-L6-v2`) and a reranker
(`ms-marco-MiniLM-L-6-v2`), about 150 MB together — from
[Hugging Face](https://huggingface.co/). Expect a one-time wait on a slow
connection; they're cached afterward, so later runs work offline.

No external LLM API key is needed to try Phileas: your MCP client's model does
the reasoning, and the embedding and reranking run on your machine.

## Connect to your AI

If you use Claude Code, `phileas init` handles this automatically.

For other MCP clients, start the server and point your client at it:

```bash
phileas serve
```

## Principles

- **Local-first** — memories stay on your machine
- **Model-agnostic** — works with any LLM via MCP
- **Natural forgetting** — irrelevant detail decays; recall favors what stays useful
- **Open** — run it yourself, read the code

For the command reference, run `phileas --help` or `phileas COMMAND --help`.

## Requirements

Python 3.11+

## License

MIT
