← All craftbooks

Spreadsheet / Financial Model

Data and analysisevalv1.0.0released 2026-06-05workflow: build-loop

Steps

Entry step: scope. Each step names the specialist role it wants; the full working prompt is expandable.

1. Scope the modelplannerentry

   lock the purpose, drivers, assumptions, and outputs

   Show working prompt

   Define the model's contract before building. Step 1: state the question the model answers and the single primary output (e.g. 'monthly net profit', '12-month cash runway'). Step 2: list the input drivers (assumptions a user changes) with a default value and unit for each. Step 3: list the calculated rows and the formula logic in words ('revenue = units * price'; 'gross margin = revenue - COGS'). Step 4: define the time axis or scenario columns if any. Step 5: write an acceptance-criteria checklist ('inputs, calcs, and outputs are visually separated', 'every assumption is a labeled editable cell', 'totals reconcile', 'changing an input flows through to the output', 'a documented sanity scenario produces the expected value'). Call write_task_note and write the driver list + formula logic + checklist to notes/scope.md.

2. Build the modeldeveloper

   implement the model with live formulas and labeled cells

   Show working prompt

   Build the model as a single interactive index.html that behaves like a spreadsheet. Step 1: lay out three clearly labeled zones — Assumptions (editable inputs), Calculations, and Outputs. Step 2: implement every calculated row with a live formula derived from the inputs (changing an input recomputes downstream values in the page). Step 3: label each cell with its name and unit; format currency/percent appropriately. Step 4: add the time/scenario columns from the scope and ensure totals sum correctly. Step 5: seed the default assumptions and show the primary output prominently. On a loop-back, fix only the named gaps. Call write_task_note with the path and the computed default output.

3. Evaluatereviewer

   Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.

   Show working prompt

   Open index.html and exercise it as a spreadsheet, verifying the math, not just the layout. Step 1: confirm inputs, calculations, and outputs are visually separated and every assumption is a labeled, editable cell with a unit. Step 2: independently compute the primary output by hand from the default assumptions and confirm the model agrees. Step 3: change one driver to a round number, re-derive the expected output by hand, and confirm the model recomputes to it live (the flow-through works). Step 4: run the documented sanity scenario and check it hits the expected value. Step 5: confirm totals reconcile (subtotals sum to totals) and no calculated value is hardcoded where it should reference an input. Write PASS/FAIL per criterion; any FAIL loops back to build.
   
   Then route — this is the whole point of the loop:
   
   - **Every criterion PASSES →** call `advance_task_step({ ref, stepId: "evaluate", next: "finish" })`.
   - **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: "evaluate", next: "build" })` to loop back. The builder fixes exactly those gaps.
   
   Never route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in.

4. Finishdeveloper

   All acceptance criteria met. Stamp a short summary and report DONE.

   Show working prompt

   Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.

Triggers

Phrases that suggest this craftbook to a crew.

• build a financial model
• make a budget spreadsheet
• unit economics model
• forecast spreadsheet
• roi calculator

Source

View this craftbook on GitHub · MIT license