Inventories tech debt and ranks it economically - scoring each item 1-3 on cost of delay (the interest rate) and leverage (the principal unblocked), multiplying into a priority index, and packaging the winners into a leadership proposal framed on business impact. Use when someone asks "which tech debt should we tackle first", "help me build a case for refactoring to leadership", "our backlog of debt is huge, where do we start", or is preparing quarterly planning. Do NOT use for finding and removing unused code - use dead-code-eliminator instead - for auditing third-party dependency risk - use dependency-risk-audit - or for planning the sprint the work lands in - use sprint-planning instead.
Click to play with sound.
---
name: Tech Debt Prioritizer
description: Inventories tech debt and ranks it economically - scoring each item 1-3 on cost of delay (the interest rate) and leverage (the principal unblocked), multiplying into a priority index, and packaging the winners into a leadership proposal framed on business impact. Use when someone asks "which tech debt should we tackle first", "help me build a case for refactoring to leadership", "our backlog of debt is huge, where do we start", or is preparing quarterly planning. Do NOT use for finding and removing unused code - use dead-code-eliminator instead - for auditing third-party dependency risk - use dependency-risk-audit - or for planning the sprint the work lands in - use sprint-planning instead.
---
# Tech Debt Prioritizer
Tech debt decisions fail when made on vibes ("this code is ugly") or recency bias (whatever hurt last week). This skill applies a lightweight economic frame - debt as a loan, where cost of delay is the interest rate you pay every sprint and leverage is the principal of future work it unblocks - so the right items get funded and the rest stay parked without guilt.
## Operating procedure
The steps run in order because scoring an unfiltered inventory is what keeps the ranking honest: filtering first means the loudest complaints survive and the quiet, expensive items never get scored.
### Step 1: gather inputs and build the inventory
Timebox to 60 minutes and cap the list at 20 items - past 20, scoring fatigue produces garbage scores. Pull candidates from three sources, one pass each:
- Friction signals: team retros and chat complaints.
- Reliability signals: incident post-mortems from the last two quarters.
- Velocity signals: areas with high cycle time or frequent reverts.
Each item gets a one-line description, the system it lives in, and the person who raised it. Do not filter at this stage; resist the urge. If an estimate is a gut call ("this costs us maybe a day a sprint"), record it as a guess.
### Step 2: score each item on two axes, 1-3 each
… load the full skill through Skill Me