Early access open

No more prompts.
Just approvals.

Press Run now on a connected repository. Tack reads the code and comes back with three ranked plans for what to build next. You pick one, approve it, and Tack opens the PR.

tackdev.com/dashboard
Tack.
acme / payments-api
Run now
Runs
Repositories
Settings
Run #128
awaiting selection
AFix session expiry on login

Smallest change that unblocks the three stalled PRs.

bugfix2 files
BRework the auth token lifecycle

Removes the whole class of expiry bugs. Wider blast radius.

refactor9 files
CAdd the weekly summary view

Highest-voted issue, but nothing is blocked on it.

feature6 files
Candidate B selectedApproveReject

The problem

Every AI tool waits.
You still decide what to ask.

v0 built your first version. Cursor helps you write code. But between those two tools, someone has to figure out what to build next. That someone is still you.

CMD

You write the prompts

Every feature, every fix — you need to describe it before any AI can touch it. That's mental overhead before a single line of code.

GIT

You scan the backlog

Which issue is most urgent? Which PR is blocking what? You're doing project management before you've written a line of code.

CTX

You re-explain the context

Every prompt starts from zero — which files matter, what shipped last week, what is blocking what. Tack carries that context for you.

How it works

Three steps.
One of them is yours.

01 — You press Run now

Tack reads your repo

Pick a connected repository in the dashboard and press Run now. Tack pulls your open issues, PRs, recent commits and diffs — read-only, nothing is written.

scanning owner/repo
├─ issues: 12 open
├─ prs: 3 open
└─ diff: +847 -203
02 — Tack proposes

Three ranked plans

One run returns up to three plan candidates for the same repo — different scope, different risk, each with the reasoning and the files it would touch.

candidates: 3
A: bugfix #42 · 2 files
B: refactor auth · 9 files
C: feature #38 · 6 files
03 — You decide

Select, approve, or reject

Compare the candidates side by side in the dashboard. Select the one that fits, approve it, and Tack implements the plan and opens the pull request.

you: select candidate B
tack: approved ✓
pull request opened
Features

Built for the day after launch.

01

GitHub-native context

Tack reads issues, PRs, commits, and code diffs — not just issue titles. It understands what's blocking what and proposes accordingly.

MVP
02

Three ranked candidates

Every run returns up to three plans with different scope and risk, ranked with the reasoning attached. Compare them side by side and pick one.

MVP
03

Approval-gated by design

Planning is read-only. Selection and approval are explicit dashboard actions, and every state change is recorded — nothing moves without you.

MVP
04

Agent execution (v2)

Approve the plan and Tack hands it off to a coding agent. PR created, reviewed, and ready for your final merge.

Coming
Why Tack

Where other tools stop,
Tack starts.

Without Tack
You write a prompt before any AI can do anything at all
🔍You manually check GitHub to decide what's urgent
🗓You plan the sprint before you can start coding
💭The AI is only as good as how well you describe the problem
With Tack
One press of Run now and Tack comes back with the plan already written
📊Tack reads your codebase and knows what's blocking what
📋You compare three ranked candidates side by side in the dashboard
Your job is judgment, not planning. Just approve what looks right.
Early access

Your next plan is
one press away.

Join the waitlist. We're onboarding solo developers and small teams first. A GitHub repository is all you need.

No credit card. No setup fee. Just fewer prompts.