AI-native product backlog

The backlog your AI agent runs.

TackPile rewrites rough product ideas into agent-ready user stories, then hands them off over MCP to Claude Code, or any other MCP agent. You prioritize. The agent ships. The board updates itself, live.

MCP-firstLive webhooksCustomizable workflowsBuilt for tiny teams

claude-code · tackpile mcp

$ claude code --next-from tackpile

› Connecting to TackPile via MCP…

› Found 4 items in "Ready for AI". Pulling priority #1.

[tackpile]TP-127 · Add OAuth loginIn Progress

› Reading product vision + agent rules…

› Drafting implementation across 7 files…

✓ Tests passing · 18/18

[tackpile]TP-127In Progress→In Dev

› PR #234 opened · linked to TP-127

[tackpile]TP-127In Dev→Done

The problem

Jira was built for sprints with humans. You ship with agents.

Your stack changed. One or two people, a Claude Code (or Cursor, or Codex) agent on autopilot, and a backlog growing faster than anyone can groom it. Tickets that read fine to a PM read like riddles to an AI. State machines designed for stand-ups don't map to an agent that opens a PR at 2am.

TackPile is a clean slate: a backlog where every item is written for a model, every state transition is a webhook, and the whole board is one MCP call away.

How it works

From rough idea to shipped, hands-off.

STEP 01
Set the vision
Drop in your product vision, terminology, tech stack, and the rules your AI agent should follow. Once. Reused on every story.
STEP 02
Pile up ideas
A sentence is enough. Tag the page or feature, mark severity, attach a screenshot. Send it to refinement.
STEP 03
AI-rewritten stories
TackPile turns each idea into a user story with acceptance criteria written for the agent, not the standup. You approve.
STEP 04
Agent ships it
Click an item or let the agent pull the top priority over MCP. States update live. Webhooks fire. The team sees progress.

Walk through a full example

What's in the box

Tools the agent reaches for. Surfaces the team trusts.

MCP-native backlog

First-class Model Context Protocol surface. Your agent can list, read, transition, and comment, just like humans do in the UI.

Idea → story refinement

AI rewrites a one-liner into a story with AC scoped to your stack and terminology. You approve before it goes live.

Customizable workflows

Replace To Do, In Progress, and Done with whatever fits your team: In Dev, In Prod, Soaking, Reverted. The agent learns it.

Live webhooks

Every transition fires a webhook. Status pages, Slack channels, your own dashboards stay in sync without polling.

Vision & rules vault

One place for your product vision, glossary, coding conventions, and agent guardrails. Injected into every refinement.

Talk to the agent on it

Each item has a thread to the agent working on it. Ask questions mid-flight. The replies update the item.

FOR PRODUCT

Drop in ideas. Get shippable stories.

Stop wordsmithing acceptance criteria for a robot. Pile up a sentence, tag the page or feature it touches, and let TackPile rewrite it into the format your agent actually follows. Approve in a click. Prioritize with a drag.

Explore product features →

FOR DEVS & AGENTS

An MCP your agent already understands.

Point Claude Code, Cursor, or any MCP-aware client at TackPile. The agent pulls the next priority, reads your product vision and rules, transitions states as it works, and posts updates back. Webhooks fire so humans see it live.

See the MCP surface →

Early access

Get on the list. Be first when we open the doors.

We're onboarding small teams shipping with AI agents. Tell us what you're building and we'll prioritize you for the beta.