PersonalOS

Your AI remembers things about you. PersonalOS lets you see, edit, and own that memory.

PersonalOS is an open-source, local-first memory and context layer for AI agents. It stores durable personal context as editable files so you can inspect, correct, version, move, or delete the memory your agent uses.

For AI builders, PersonalOS is a practical reference pattern for user-owned memory: agent-agnostic policy, runtime adapters, skill routing, and review-before-save durable context.

It is for people who like using AI, but do not like starting from zero every time — or trusting a hidden cloud memory they cannot inspect.

---

The problem

AI memory is becoming useful. It is also becoming invisible.

You may not know:

• what your assistant remembered;
• where that memory lives;
• how to correct it;
• how to delete it;
• whether you can take it to another agent.

PersonalOS changes that by keeping your working context in files you control.

---

What PersonalOS gives you

• A private memory folder your agent can read and update with your approval.
• A profile for how you think, communicate, write, and work.
• Review before memory is saved — no silent durable personalization.
• A workspace that can move with you across agents and machines.
• An open-source foundation you can inspect, adapt, or version privately.

---

What you can use it for

PersonalOS gives your agent one private memory layer that different skills can use across different kinds of work:

• Think through decisions with context about how you reason. Me / default thinking mode
• Get coached with memory of your goals, patterns, and commitments. Executive Coach
• Design systems using your known architectural preferences and constraints. Technical Architect
• Prepare career stories from your real work and decisions. Job Search
• Draft in your voice instead of sounding like a generic assistant. Me / Social
• Manage recurring projects without rebuilding the backstory every session. Assistant

Skills are reusable ways your agent can behave for different kinds of work. They become more useful because they draw from the same reviewed memory, profile, and context. You do not need to choose a skill first; start by asking for the outcome you want.

---

What it feels like

You: Help me think through this decision.

Agent: I remember that you prefer small reversible moves when the stakes are unclear.
Want me to use that lens here?

And if the agent learns something new:

Agent: This seems worth remembering. Should I save it?

You stay in control of what becomes durable memory.

---

Start simple

You do not need Git to start. If you are evaluating the project from GitHub, the core loop is:

conversation -> proposed memory -> user review -> Markdown files -> future agent context

Try it locally:

Download PersonalOS
→ put the folder somewhere safe
→ open that folder in your AI agent
→ run the start command
→ review what the agent wants to remember

You will need Python installed to run the setup commands. If you are not sure, the getting started guide shows how to check.

You will run two short commands in a terminal. If you have never done that, copy exactly what is shown in the getting started guide.

Start here:

python scripts/personal-os.py onboard
python scripts/personal-os.py start

For the full first-use walkthrough, read GETTING_STARTED.md.

Privacy note: your files stay in the folder you choose. If you put that folder in OneDrive, Google Drive, iCloud Drive, or Dropbox, that provider may sync it like any other folder. Read PRIVACY.md before storing sensitive context.

---

How PersonalOS is different

Other tools often give you...	PersonalOS gives you...
Memory hidden inside an app	Memory as files you can open and edit
A new context in every tool	A reusable context layer for your agents
Silent personalization	Review before memory becomes durable
One company owning the experience	An open-source folder you control
Task automation only	Continuity in how you think and work

---

Who should try it?

Try PersonalOS if you:

• use AI often enough that re-explaining yourself is annoying;
• want an assistant that understands your working style;
• care about privacy and inspectable memory;
• want to use AI for thinking, writing, coaching, planning, product work, or career work;
• like the idea of your agent's memory being something you can review.

---

If you are technical

PersonalOS can also be used as a private Git repo.

That gives you:

• version history for your personal operating layer;
• diffs for memory changes;
• branches for experiments;
• safer contribution workflow for improving the open-source project.

Keep any repo containing personal memory private unless you intentionally redact it.

Technical docs start at docs/index.md. AI agent builders should start with docs/technical-overview.md.

---

Where to go next

• First time? Read GETTING_STARTED.md.
• Worried about privacy? Read PRIVACY.md.
• Looking for the full documentation map? Read docs/index.md.
• Building agents or researching AI memory? Read docs/technical-overview.md.
• Want launch/community assets? See docs/community-launch-assets.md.
• Want the product website? See site/, intended for https://www.movetheneedle.info/personal-os/.
• Configuring GitHub discovery? Recommended repository topics: ai-agents, personal-ai, ai-memory, local-first, agentic-ai, open-source, knowledge-management, llm-tools.
• Need technical depth? The documentation map points to docs/principles.md, docs/architecture.md, docs/file-ownership.md, docs/release-qa.md, and docs/reorganization-proposal.md.

---

Current skills

PersonalOS currently includes skills for general thinking, executive coaching, technical architecture, assistant tasks, social content, job search, and creating new skills.

You do not need to learn them all now. Start with the outcome you want, and PersonalOS can route to the right behavior.