hobkin
Get early access

Your best pal for real work.

hobkin takes the friction out of your day. It's an AI agent in your terminal that learns how you work and handles the small tasks, so you can focus on the creative ones.

Get early access See what it does

Works with Pi, Claude Code and Codex.

Schedules

Start your day knowing what changed.

Tell hobkin in Chat what you want checked and when, and it handles the rest. Each check runs on the model you choose, and a report is waiting when it's done.

The Schedules tab, with example checks. hobkin works while it's open, and on a Mac it keeps the machine awake while the screen sleeps.

Loop engineering

Every check keeps notes on what it found, what's still open and what it already reported, so each run builds on the last.

Only what needs your attention

Findings worth your time appear in Chat. When nothing has changed, the report stays in Reports and you aren't interrupted.

Clear when something fails

After two failed runs in a row, hobkin holds the check, finds the cause and brings you a plan. Nothing changes until you approve it.

Work mode

Steady progress on the work that keeps getting postponed.

Type /work, choose a folder and pick the projects you want help with. Agents work through each one in small, reviewable changes, so your backlog keeps moving while you focus on your own priorities.

  • Security
  • Tests
  • Features
  • Bugs

The Work Mode tab, with example projects. In Herdr you can follow each agent in its own pane, on Pi, Claude Code or Codex.

Every task has a reason

An agent can't file a vague idea. It has to show the evidence, say who benefits, name the first change and explain how it will prove the work is done.

You review and merge

Every task starts with a test or check that fails and ends on its own branch, with a description ready for your review. Agents never commit to your main branch, and nothing is pushed unless you ask.

Your subscription is protected

Work mode has a daily token limit, keeps 30% of every usage window for your chat and checks, and paces itself across the week.

Memory and learning

It learns how you work.

Remembers your preferences

Tell hobkin once how you like things done and every agent follows it, in chat, checks and work mode. What it remembers is stored on your machine, and agents search it before repeating research.

Learns from results

With learning on, hobkin looks at how work turns out: which reports you open, which proposals you accept and which PRs you merge. It turns that into a few short lessons for every agent, and working them out uses no model and no tokens.

What the lessons look like
learned: advice only, grants nothing
-site-audit blocked x5: do the rest, name what stopped it
-news-digest unread x5: lead with the change, be short
+price-watch used x5: keep this approach

Early access

Be among the first to try hobkin.

hobkin is in private release for now. Leave your email and I'll write to you once, when it's ready.