{"slug": "show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with", "title": "Show HN: Phileas – Local-first long-term memory for the AI you chat with", "summary": "Phileas, a local-first long-term memory layer for AI chatbots, launched on Hacker News. The open-source tool uses MCP to let any LLM read and write persistent context across sessions, running two small models locally on the user's machine.", "body_md": "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.\n\n**Python 3.11 or newer.****An MCP client.** Phileas exposes its memory over MCP. The setup wizard wires it into[Claude Code](https://docs.claude.com/en/docs/claude-code)automatically; any other MCP client (a GPT or Ollama front-end, and so on) connects to`phileas serve`\n\n.**A few hundred MB of disk and one download.** Phileas runs two small models locally (about 150 MB) and depends on PyTorch. The steps below keep that download lean.\n\n```\npython3 -m venv ~/.venvs/phileas # an isolated environment\nsource ~/.venvs/phileas/bin/activate\n\npip install torch --index-url https://download.pytorch.org/whl/cpu\npip install phileas-memory\n\nphileas init\n```\n\nPhileas uses PyTorch only to run two small models (an embedding model and a reranker) on the CPU, so the first install line fetches PyTorch's lean CPU build. That keeps the download to a few hundred MB; without it, `pip`\n\npulls the multi-gigabyte CUDA build on Linux. On macOS and Windows the CPU build is already the default, so that line is harmless there too.\n\n`phileas init`\n\nis the setup wizard: it chooses where memories live, connects Phileas to Claude Code, downloads the models, and starts the background daemon. Restart Claude Code afterward so it picks up the memory tools.\n\nThe MCP server is launched by its full path, so Claude Code finds it whether or not the venv is active. To run `phileas`\n\ncommands yourself (such as `phileas status`\n\n), activate the venv first.\n\nThe first run downloads two models from [Hugging Face](https://huggingface.co/) that then run locally: an embedding model (`all-MiniLM-L6-v2`\n\n) and a reranker (`ms-marco-MiniLM-L-6-v2`\n\n), about 150 MB together. They are cached after the first download, so later runs work offline. No external LLM API key is needed: your MCP client's model does the reasoning, while embedding and reranking run on your machine.\n\n`phileas init`\n\nwires up Claude Code. Any other MCP client connects to the same server: `phileas serve`\n\nspeaks MCP over stdio, so register it the way that client registers a stdio server.\n\nMost clients (Cursor, Antigravity, and others) read a JSON config with the same shape. Add Phileas to its `mcpServers`\n\nmap:\n\n```\n{\n  \"mcpServers\": {\n    \"phileas\": {\n      \"type\": \"stdio\",\n      \"command\": \"/absolute/path/to/phileas\",\n      \"args\": [\"serve\"]\n    }\n  }\n}\n```\n\nUse the absolute path to the `phileas`\n\nexecutable (run `command -v phileas`\n\nto find it), since the client launches it without your shell's PATH or an active venv. To point a client at a named profile, add `\"env\": {\"PHILEAS_PROFILE\": \"<name>\"}`\n\nto the entry.\n\nThe tools work in any client, and the server ships usage guidance that every MCP client receives on connect. The extra layer that makes memory feel automatic (recall before answering, memorize when something worth keeping comes up, the query shapes that retrieve well) lives in a skill file. `phileas init`\n\ninstalls it for Claude Code at `~/.claude/skills/phileas/SKILL.md`\n\n; its text uses bare tool names so it carries to any client. If your client has a rules or skills mechanism, put that file's content into it.\n\n**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.\n\nFor the full command list, run `phileas --help`\n\nor `phileas COMMAND --help`\n\n.\n\nMIT", "url": "https://wpnews.pro/news/show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with", "canonical_source": "https://github.com/alexajuno/phileas", "published_at": "2026-06-22 00:07:18+00:00", "updated_at": "2026-06-22 00:26:10.796504+00:00", "lang": "en", "topics": ["large-language-models", "ai-tools", "ai-agents", "developer-tools"], "entities": ["Phileas", "MCP", "Claude Code", "PyTorch", "Hugging Face", "all-MiniLM-L6-v2", "ms-marco-MiniLM-L-6-v2", "Cursor"], "alternates": {"html": "https://wpnews.pro/news/show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with", "markdown": "https://wpnews.pro/news/show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with.md", "text": "https://wpnews.pro/news/show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with.txt", "jsonld": "https://wpnews.pro/news/show-hn-phileas-local-first-long-term-memory-for-the-ai-you-chat-with.jsonld"}}