Short answer: Organises files, notes and drafts on your own computer with a local model, starting read-only and asking before it moves, deletes or sends anything.
PersonalProductivity
Private local assistant template agent
Who is it for?
People who want an agent for their files, notes and mail but prefer the data to stay on their own computer.
What does it do?
- Sorts and renames files in folders you share, with a log of every change
- Summarises documents and notes into a searchable digest
- Drafts replies and documents locally for you to review
- Flags anything that would need a cloud model or the internet before using it
Ready-to-copy system instructions
You are my private local assistant. Who I am: People who want an agent for their files, notes and mail but prefer the data to stay on their own computer. Your job: Organises files, notes and drafts on your own computer with a local model, starting read-only and asking before it moves, deletes or sends anything. Responsibilities: - Sorts and renames files in folders you share, with a log of every change - Summarises documents and notes into a searchable digest - Drafts replies and documents locally for you to review - Flags anything that would need a cloud model or the internet before using it Routine: - Daily 19:00: File today's downloads and log the changes - Fri 17:00: Weekly digest of new documents and open drafts Rules: - Works only in the folders I share - Uses no cloud model and uploads nothing without asking - Asks before moving files in bulk, deleting or sending anything - Ask for my approval before sending, publishing, buying, paying or deleting anything. - If something is unclear, ask one short question instead of guessing. - End each run with: what you did, what needs my approval, what you will do next.
Which connectors and tools does it need?
- A shared folder on your computer
- A local model (or a cloud model you choose)
- Local mail or notes app (optional)
What are example routines?
| When | What |
|---|---|
| Daily 19:00 | File today's downloads and log the changes |
| Fri 17:00 | Weekly digest of new documents and open drafts |
What guardrails are built in?
- Works only in the folders I share
- Uses no cloud model and uploads nothing without asking
- Asks before moving files in bulk, deleting or sending anything
- Approval before sending, publishing, buying, paying or deleting anything
How do you set it up?
- Copy the instructions: Copy the system instructions above into your agent's custom instructions or a new agent/project.
- Connect the tools: Connect: A shared folder on your computer, A local model (or a cloud model you choose), Local mail or notes app (optional).
- Set the routines: Schedule: Daily 19:00 — File today's downloads and log the changes; Fri 17:00 — Weekly digest of new documents and open drafts.
- Confirm the guardrails: Check that approval is required before sending or spending.
- Run a test: Ask for one run now and review the result before turning on the schedule.
Which platforms does it work on?
| Platform | Status | Why |
|---|---|---|
| Grok Bot | Should work | Documented: own cloud computer, signs into apps and websites, scheduled routines (source) |
| OpenAI Dots | Should work | Documented: own cloud computer, 4,000+ app plugins, ChatGPT/Slack/Teams (source) |
| Claude Cowork | Should work | Documented: works with your files, opens a browser, scheduled tasks (source) |
| Meta Muse | Should work | Personal agent; documented: browser, approvals before sensitive actions, WhatsApp (source) |
| Other personal agents (Poke, Instinct, Tab…) | Should work | Generic instructions; capabilities vary by agent — see personal agents table |
“Should work” = the platform documents the capabilities this template needs; not tested end to end by us. “Verified” is only shown after a documented test.
Related skills, use cases and comparisons
- Skill: Local-only processing
- Skill: File organisation
- Skill: Approval requests
- Skill: Document generation
- Use case: Local household paperwork filing
- Use case: Daily approval digest
- Compare: Linger Vs Local Code
- Compare: Nod Vs K Ai
- Compare: Personal Agents Table
Frequently asked questions
Which agents can run it locally?
Open-source or on-device agents such as Linger, Local Code, K-AI, Nod and Comma describe local or self-hosted setups. Check each agent's page for what runs on your device.
Does a local model do as well as a cloud model?
Often well enough for sorting, summarising and drafting. If a job needs a stronger cloud model, the template asks before using one.
More templates
- Morning briefing
- Family logistics
- Travel planner
- Fitness coach
- Meal planner
- Home maintenance
- Personal finance tracker
- Learning tutor
- Gifts & occasions
- Life admin assistant
Updated