daily ops

Open source · Local first

A plan for the
time you have.

Turn a long task list into a manageable day. See what fits, keep the rest visible, and pick up where you left off.

Try a day Make it yours

No account or API key for the local tool.
Use it on its own, or with your AI assistant.

01 / PLAN

Make the tradeoffs visible.

Give your tasks estimates and choose your time budget. Get a clear shortlist, with reasons for what stays out.

02 / ADAPT

Leave room for real life.

Plans change. Reduce your available time, update a task, and plan again. Unfinished work stays intact.

03 / KEEP

Take your context with you.

Your tasks live in portable files. Resume in a new session, switch assistants, or work without one.

A working example

What fits your day?

Pick a fictional routine. Change the minutes.
Watch the plan make the tradeoffs explicit.

Choose a fictional routine
Fictional data · In your browser

Your day, on your terms

Between classes

A lab deadline, some reading, and more good intentions than minutes.

Time budget

After classes, meetings, and other fixed commitments.

Room for a break or an interruption.

Task budget80 min

Example date: 7 October 2026.
Changes here are not saved.

Your shortlist

75 min planned

5 min unallocated

SelectedUnallocatedReserve

Start here 2

  • Enable JavaScript to explore this plan.

For later 2

    Waiting 0

      How it chooses: due dates first, then priority, then task ID. Whole tasks must fit the remaining budget. Waiting tasks stay visible.

      This demo runs a fixed planning rule. It does not call an AI model or send your inputs anywhere.

      New in 1.1 · Review your decisions

      See what a change would change.

      A real report from the local tool.
      Four fictional tasks. Two decisions. One clear preview.

      01 / UNDERSTAND

      A full day can still miss a deadline.

      Our example fills 45 minutes, but the due proposal is too large and the room arrangement is waiting. Both decisions stay visible beside the facts.

      Open the interactive plan
      02 / PROPOSE

      Keep the choice in your hands.

      Try changing an estimate or clearing a blocker. Each response starts unreviewed. Download a proposal and notes to keep your choices. Opening or editing a report never updates saved tasks.

      The worksheet keeps drafts only until you reload. Downloads contain this fictional example.

      03 / PREVIEW

      Know what moves before applying.

      If the proposal really needs 25 minutes and the venue has replied, preview both changes. The two due tasks now fit in 40 minutes. The other work stays visible for later.

      See the before-and-after preview

      These reports are generated by the same Python runtime in every release. The local validator catches stale revisions and changed plan facts. Read what changed and why.

      Small by design

      Your files.
      Your assistant.
      Your next step.

      The local tool needs Python 3.10 or newer. It has no runtime dependencies and makes no network requests. Start from the source, or follow the guide for your assistant.

      Read the setup guide
      01

      Run it locally

      Capture tasks, plan within a budget, record outcomes, and open a clear HTML report. Your JSON workspace stays on your computer.

      python3 skills/daily-ops/scripts/run.py --help
      02

      Bring an assistant

      Use the skill with a host that can run Python and access your workspace. Ask it to clarify a task or help you start one. The same local planner handles the arithmetic.

      03

      Use an ordinary chat

      Share a task snapshot and ask for a proposed plan. Review the changes and save the snapshot yourself. Without a runtime, automatic validation and persistence are unavailable.

      A few honest answers

      Know what to expect.

      Is Daily Ops an AI scheduler?

      The local planner uses a documented rule to order tasks and fit their estimates into a budget. It does not read a calendar or find an optimal schedule. An optional assistant can help turn a vague intention into a clearer task, but its suggestions still need your judgment.

      Does my task list leave my computer?

      The core does not upload tasks or collect telemetry. If you choose to paste a snapshot into Claude, ChatGPT, or another hosted assistant, you share those contents with that service under its terms. Local storage does not mean local AI inference.

      Will it save me time?

      That has not been established. The demo shows budget arithmetic and visible tradeoffs. The project tests those behaviors; it does not claim measured productivity gains. Try the workflow and judge whether the decisions it helps you make are worth the effort.

      Can I switch assistants or stop using it?

      Yes. Task state is stored in a documented JSON format with stable IDs, and reports are ordinary HTML or Markdown. Export and keep your files. The core works without an AI account; AI host access, fees, and capabilities are separate.