Sprintfy

Built on WebMCP

Your browser has WebMCP switched off, so the tools on this page are dormant. Everything below still works — turn the flag on and reload to watch them arrive.

What would you like to do?

Your agent is waiting on you

Measure before rebuilding

The brief says drop-off is around 40%, but nothing records it per step. I put a short instrumentation sprint in front and re-sized the route collapse.

+3~14 operations
  1. +

    sprintSprint 0 — Instrument first

    Know the drop-off rate per step before changing anything.

  2. +

    taskAdd funnel events for each checkout step

    Sprint 0 — Instrument first (new) · 4h

  3. +

    requirementCheckout drop-off is measurable per step

    high priority

  4. ~

    taskCollapse the three checkout steps into one route

    • statusBacklog→To do
    • estimate8h→12h
Nothing is written yet.RejectApprove 4
01

It reads the room

get_project_context returns the plan and the human's live view — the open sprint, the active filter. The agent proposes into a context, not a vacuum.

02

It writes a diff

Every write tool authors a change set. Additions, field-level changes, removals. A sprint and the twelve tasks inside it arrive as one reviewable unit.

03

You answer, in-band

apply_pending_changes blocks on your decision. Approve and it lands in one transaction. Reject and your reason goes back as the tool result.

The tool surface

  • get_project_contextThe plan, and what the human is looking at right now.
  • propose_planA rough brief in; requirements, sprints, tasks and estimates out.
  • propose_changesRetitle, re-estimate, move sprints, reprioritise, delete.
  • apply_pending_changesAsks the human. Waits. Returns what they decided.
  • discard_pending_changesWithdraw a proposal that missed.
  • focusMove their view, and ring the thing you're talking about.

The set changes with the page. The projects list offers list_projects and create_project instead — which is what toolchange is for.

Turning WebMCP on

It needs desktop Chrome. The badge in the header turns on and lists the live tools.

  1. 1chrome://flags/#enable-webmcp-testing
  2. 2Set it to Enabled, then relaunch Chrome.
  3. 3Reload this page.

To drive it with ChatGPT, use its Chrome extension in that same browser — or skip the flag entirely and open this page in the ChatGPT desktop app, in ChatGPT mode: its built-in browser exposes WebMCP on its own. Codex mode in that same app does not, so if the tools are dormant there, that is why.