Factory — Wireframe to Backlog | Free for Claude Code

From wireframe to a backlog you can quote against.

Factory is a set of Claude Code skills derived from real shipped sprints. Upload a wireframe and it builds the full backlog — including the auth, payments, billing, and webhook stories most estimates forget — then sizes every story against real delivery actuals. Plus change-request impact assessments and sprint reviews that compile to PPTX. Free on the official Anthropic marketplace.

/plugin install factory@newecom-ai/factory-skills

Works in Claude Code, Cursor, Codex CLI, and Gemini CLI.

Backlogs that don't lie about infrastructure.

Most estimates don't blow up because a story was too small on paper — they blow up because the story was never written down. Auth needs password reset, email verification, and session handling. Payments need webhooks, retries, and reconciliation. Factory reads the wireframe and writes those stories in, before they turn into change requests.

And the stories that do make the estimate get sized against real delivery actuals. On a real SaaS billing build, the billing epic was estimated at 14 hours. It took 64 — a 4.6× miss, the kind of number that turns a 50% margin into a loss-leader. Factory carries the correction factor we derived from that.

Infrastructure — auth, payments, billing, webhooks — is where fixed-bid projects quietly lose money: from the stories that get forgotten and the ones that get underestimated. This is the data competitors can't replicate without comparable delivery experience. We're shipping it free.

Five skills, one calibrated backlog

wireframe-to-backlog

From wireframes to a calibrated backlog

Upload wireframes, get a calibrated backlog with hours, epic grouping, sprint assignment, and discovery gaps. 5-pass method.

change-request-impact

Change-request impact, sized and documented

Classify, size, run absorbability tests. Generates the internal record AND the client-facing email. Response SLA built in.

brand-spec-generator

A full brand spec in ten minutes

10-minute brief → full WCAG 2.1 AA HTML spec, 13 sections, .cursorrules block ready to paste.

sprint-review

Sprint reviews that compile themselves

Compiles a 10-slide PPTX from sprint metrics + handoff doc.

cursor-story-prompt

Backlog story to Cursor prompt

Backlog story → Cursor implementation prompt with project invariants. Stops the "what about TypeScript strict mode" back-and-forth before it starts.

For people who sign contracts and ship code

  • Fractional PMs estimating fixed-bid projects
  • Indie dev shops who lose money on infrastructure surprises
  • Solo founders pricing their first MVP build
  • Anyone running discovery → estimate → sprint loops weekly

Not for: PMs who only manage roadmaps. This is for people who sign contracts and ship code.

Factory MCP — in development Not built yet

The free skills ship with calibration data as of June 2026. They work forever — but the calibration doesn't get fresher. A hosted version would fix that. We haven't built it, and we're deciding whether to. If these are the pieces you'd want, tell us and we'll build them:

  • Live calibration, refreshed monthly from real shipped sprints
  • Persistent project state — no re-typing your stack on every call
  • Linear, Jira, and GitHub sync
  • Multi-project workspaces

No signup, no product to buy yet — this is us checking whether it's worth building. The five skills above are free and work today either way. Tell us what you'd want:

These skills came out of client delivery.

Factory is the system NewEcom.AI uses to ship client projects — the estimates, the change-request process, and the sprint reviews are the same artifacts our clients receive. If you'd rather we ran the build than run it yourself, that's the day job.

  • 11–18stories shipped per sprint
  • 46–62minmedian pull-request duration
  • ~100test definitions per story

Why we open-sourced this

NewEcom.AI is an AI-assisted product development studio. Factory is the skill system we use to ship client projects with consistent velocity, quality, and commercial transparency. We open-sourced the skills because the bottleneck for most PMs and dev shops isn't access to AI — it's access to calibrated AI.