Screenshot in · finished task out

Hand your agent a URL. Get the work back done.

Mark up a screenshot, publish the list as a private URL, and your agent works through it one task at a time.

dothesetasks.com/projects/hero-audit Draft dothesetasks.com/projects/hero-audit Draft dothesetasks.com/projects/hero-audit Draft dothesetasks.com/x/lp_9f3ax2 Published dothesetasks.com/x/lp_9f3ax2/next.md Published dothesetasks.com/x/lp_9f3ax2 Closed

New task

AI Enhance · 5 cr

Hero H1 overflows below 380px

and pushes the CTA off-screen.

viewport-380px.webp nav-overflow.webp

Ship your next
idea this week

Audit #3

Group · Mobile layout

#412 Cap the hero font scale below 380px
#413 Keep the primary CTA above the fold
#414 Remove the horizontal scroll at 360px
#415 Re-test the nav drawer after the fix

Brief URL

dothesetasks.com/x/lp_9f3ax2 Copy Copied

/index.md: the whole job, so the agent can see the shape of it
/next.md: one unit at a time, so nothing gets skimmed

$ curl -s dothesetasks.com/x/lp_9f3ax2/next.md

200 Group: Mobile layout · 4 tasks

Cap the hero font scale below 380px
Keep the primary CTA above the fold
Remove the horizontal scroll at 360px
Re-test the nav drawer after the fix

POST /x/lp_9f3ax2/tasks/415/completed → 200

4 of 4 tasks complete

dothesetasks.com/x/lp_9f3ax2

The list closes and the brief unpublishes itself — the URL stops answering.

Capture the issue — type what's wrong, drop the screenshots that prove it.

Mark up the screenshot — draw an arrow, circle the bug, pick a colour.

Build the list — every note becomes one task with its own done-when.

Publish a private URL — the list becomes a brief URL you can hand over.

Hand the URL to your agent — it pulls one task, builds it, then calls back.

It closes itself — the last callback lands and the brief unpublishes.

Agent-ready task briefs

Turn screenshots into tasks your agent actually finishes.

DoTheseTasks turns notes and screenshots into a private brief your coding agent pulls one task at a time — so it builds each one instead of skimming a long list and stubbing half of it.


              $ curl -s dothesetasks.com/x/lp_9f3ax2/next.md

200 OK

# Group: Mobile layout (2 tasks)

Finish every task in this unit, then request the next one.

---

## Task ID: 412

Fix the oversized hero on mobile. The H1 overflows below 380px and pushes the CTA off-screen. Cap the font scale and keep the primary button above the fold. Done when: CTA visible at 360px with no horizontal scroll.

![viewport-380px](dothesetasks.com/i/47fa.png)

Task completion URL: dothesetasks.com/x/lp_9f3ax2/tasks/412/completed

Your agent pulls one unit at a time — not all 27 tasks at once. It builds, calls back, then asks for the next.

How it works

Capture once. Hand off a URL. Watch it close.

01
Capture and enhance

Paste one issue per task and drop the screenshots that prove it — from your desktop or the Chrome extension. AI Enhance rewrites rough notes into implementation-ready tasks with clear acceptance criteria.
02
Group and publish

One click groups related tasks into units of work. Publish to a private brief URL: an index of the whole job plus a feed that serves one unit at a time.
03
Agent pulls and closes

Your agent fetches the next unit, builds it, and hits a completion callback — then asks for the next. The brief tracks progress and closes itself when the work lands.

Why it works

What makes a brief agent-ready.

One unit at a time

The brief is an index plus a next feed. The agent pulls a single task or group, finishes it, and asks for more — so context stays small and nothing gets stubbed.

AI Enhance

Turns a rough note and a screenshot into an implementation-ready task — objective, concrete steps, and a "done when" checklist the agent can verify against.

Capture from any page

The Chrome extension grabs a screenshot straight from the tab you're on and drops it into a task — no context-switch, no file wrangling.

Completion callbacks

Agents mark each task done through a callback URL as they go. You verify one task at a time against its criteria instead of auditing one giant diff.

Hand the URL to the agent you already use — Codex · Claude Code · Cursor · Windsurf

Write the brief once. Let the agent build from it.

Your next audit can ship as a URL instead of a conversation.

Start free Log in

Free to start · no credit card · Premium is $28/month for every Wynter app