Local-first Personal Model Runtime for macOS
Persome is a local-first macOS Runtime that collects focused cross-app context, builds an inspectable model of one real person, and exposes it to trusted MCP clients.

The image is a real Runtime screenshot generated from synthetic fixtures. Open
the authenticated real-data viewer with persome model open after installation.
Star Persome on GitHub to follow releases and help prioritize the next MCP integrations.
git clone https://github.com/Intuition-Lab/personal-model.git
cd personal-model
uv run python scripts/sample_demo.py
bash install.sh
persome onboard
persome status
persome llm status
persome ocr status
persome doctor
persome model open
The installer already runs onboarding in an interactive macOS session and
leaves the Runtime running; the explicit persome onboard above is an
idempotent recheck, not a prerequisite stop/start cycle. It requests
Accessibility for the actual source-versioned AX helper and watcher, applies
the durable OCR policy, proves the final Runtime owner and generation, and
returns the receipt appropriate to daemon, ingest, paused, or locked mode.
Update an existing installation from any directory with persome update. The
command builds an inactive candidate, atomically exchanges it with the active
virtualenv, preserves local configuration and personal data, and commits only
after the same mode-aware onboarding proof. An interrupted update exchanges the
old Runtime back before restoring its lifecycle owner.
Persome keeps data under ~/.persome, uses local AX capture by default, offers
optional on-device OCR, and serves streamable HTTP MCP at
http://127.0.0.1:8742/mcp behind an owner-local bearer. Normal client
installers use stdio and do not copy that credential.
Persome is the project and Runtime name. Personome is the research term for the learned personal model. Publication artifacts and research benchmarks are separate works with independent licenses.