Favro

Promptfree, for everyone, and the only version there is

Copy the prompt for Claude CodeCopy the prompt for CodexCopy the prompt for Cursor

Copy

Build me the multi-level planning tool that replaces Favro.

The idea worth copying: **one card, appearing on several boards, with a different column on each.** Most tools force a card to belong to one board, which is why every organisation ends up copying cards upward and keeping the copies in step by hand.

STACK
- Node 20+ with Fastify, server-rendered HTML with a little vanilla JS
- SQLite through better-sqlite3, WAL mode, FTS5 across cards
- Caddy in front

THE DATA MODEL, WHICH IS THE WHOLE DESIGN
- cards: id, title, body_md, created_at, archived_at — **the card exists once and belongs to no board**
- boards: id, name, kind, columns_json
- placements: id, card_id, board_id, column, position, added_at, removed_at — **append-only; a card leaving a board is a removal row, not a delete**
- relations: id, parent_card_id, child_card_id — an epic is a card with children, not a different kind of thing
- assignments, estimates, tags: all on the card, shared everywhere it appears
- changes: id, card_id, field, before, after, actor, at

WHAT THIS MAKES POSSIBLE
- A team board with the columns that team uses, and a portfolio board with three columns, showing the same cards. Moving a card on one does not move it on the other, and **the card's content is one thing so it cannot diverge**
- Roll-up: a parent card shows the state of its children automatically. Nobody types a status
- A card can be planned at portfolio level and executed on two team boards at once, which is what actually happens in organisations and what most tools cannot express

THE RULES THAT KEEP IT COHERENT
- Archiving a card removes it from every board at once and keeps the placements history
- **Deleting is not offered.** Archive with a reason. A deleted card takes its history and everybody's context with it
- A card must always be reachable: a search that finds it regardless of which boards it sits on, because the failure mode of this model is a card on no board that nobody can find

THE BOARD
- Drag between columns, keyboard reachable, positions as sparse integers so a move is one write
- Column limits with a visible warning rather than a block
- Filters by tag, assignee and age, saved per person

REPORTING
- Time in column, from the placements and changes history, per board
- Cards with no movement for a fortnight, per board — **the single most useful report in any planning tool**
- Throughput per week, and cycle time as a distribution

WHAT MATTERS MOST
The card existing independently of boards, and placements being append-only. Together they remove the copying that every other tool forces, which is the entire reason this product exists.