Change Request Log & Impact Assessment: No Scope Creep Without a Price Tag
Capture every scope change in one log, let AI draft a structured impact assessment (effort, cost, schedule, risk, affected deliverables) against your versioned baseline, then have the PM review and complete it before anything is approved — so nothing slips into scope with its true cost hidden.
A web tool where anyone can log a change request, AI drafts a structured impact assessment (effort, cost, schedule, risk, affected deliverables) measured against your versioned baseline, the PM reviews and edits it, and only an approved, fully-assessed CR moves to a decision — with everything exported as a clean CSV.
Before you start
- A Supabase account (free)
- A Vercel account (free)
- A Resend account (free)
- Your current project plan/baseline as a CSV or sheet (tasks, dates, budget)
- A change-type / impact-rules reference as a CSV or sheet
- Claude Code or any AI coding agent
The problem this kills
Scope creep almost never arrives as a big, obvious decision. It arrives as "can we just add this one little thing," dropped into a hallway conversation, a Slack thread, or the bottom of a status email. Someone says yes to be helpful. No one writes down what it actually costs. Three weeks later the timeline is blown, the budget is gone, and nobody can point to the moment it happened.
The fix everybody knows is a change-control process: log the change, assess its real impact on effort, cost, schedule, and risk, and don't action it until someone with authority signs off. The reason it gets skipped is that doing the impact assessment by hand is slow and fiddly — you have to find every task the change touches, re-do the math against the current plan, and write it all up while five other fires are burning. So people wave it through instead.
This tool removes the friction. The moment a change request is logged, the AI drafts the impact assessment for you against a versioned snapshot of your plan. The PM's job shrinks from "build the assessment from scratch" to "review, correct, and complete the draft." Suddenly the true cost is always on the table before anyone says yes — and you didn't have to become a developer to make that happen.
What you'll build
A small, private web app for your project team where:
- Anyone on the team can log a change request with a title, description, who asked for it, and why.
- The tool catches duplicates automatically — both on an explicit CR id and on the same project plus a normalized title — so the same request can't quietly get assessed twice.
- The AI drafts a structured impact assessment: effort, cost, schedule slip, risk rating, and the specific tasks and deliverables affected, all measured against the versioned baseline the CR is tied to.
- The PM reviews, edits, and completes that draft — nothing is taken as final just because the AI wrote it.
- A fully assessed CR moves into a ready-for-decision state and is handed off for formal approval; it is never actioned until then.
- Everything exports as a clean CSV in the columns your PMO or PPM tool expects.
What's inside the Implementation Plan
The plan is a single markdown file you paste into an AI coding agent (like Claude Code). It walks the agent through building the whole tool, step by step, with a ready-to-copy prompt at the end of each step.
Crucially, the plan opens by interviewing you about your business — your current change process, the systems and spreadsheets you use, the real fields and naming in your plan and rules, your typical and peak change volumes, your exact approval rules, and your messiest edge cases. It then reflects a short tailored spec back to you for a thumbs-up, and shapes the data model, the impact math, and every later build step around your answers. You get a tool that fits how your project actually runs — not a generic template you have to bend yourself around.
Inside you'll find: the discovery interview, the definition of done, an architecture diagram, the build steps (data model, baseline versioning, the logging form with duplicate guards, the AI impact drafter, the PM review-and-approve gate, the CSV export), a verification checklist, and a no-API fallback so you can build the whole thing today from sheets alone.
The governance it includes (this is the point)
This isn't a toy. The plan builds in the controls that make a change log trustworthy:
- Login so only your team can see or touch it.
- Row-level security so each organization only ever sees its own projects and change requests.
- A complete audit trail — who logged what, who edited the assessment, who approved, and when.
- A hard human-in-the-loop approval gate: the AI only ever drafts; the PM reviews and completes the assessment, and a CR is not actioned until it is formally approved.
- Duplicate guards on CR id and on (project + normalized title) so the same change can't be processed twice.
- A versioned baseline so every impact assessment is measured against a known, frozen snapshot of the plan — not a moving target.
Who it's for
Project managers, business analysts, delivery leads, and sponsors who are tired of watching scope creep in unmeasured — and who want a clean, defensible record of every change and what it really cost, without hiring a developer or buying a heavyweight PPM suite.
You've got this — paste the first prompt and let the agent interview you.