Your tech debt backlog is sprawling. Fifty items, some critical, some nice to eventually fix, and you do not know which ones are actually slowing you down. So you either ignore them and watch velocity decline, or you pour too much sprint time into debt and miss features customers want. The Tech Debt Analyzer agent quantifies the impact. Every Monday at 10 AM it tells you which debt is actually costing velocity and in what order to fix it: the auth refactor is painful but affects 5 percent of stories, the legacy data layer affects 40 percent of feature work, so fix the data layer first.
How it works
The agent pulls from three sources and applies scoring. Backlog analysis reads every debt ticket in Jira or Linear (description, status, estimated effort, age, related features). Velocity correlation analyzes which stories touched which debt areas and how much slower they moved: if stories touching the auth system take 40 percent longer, that is signal. Feature impact checks which upcoming features will hit the most debt: if your top 3 prioritized opportunities all require fixing the data layer, that debt is urgent.
The output is a list scored by (impact times urgency) divided by effort, recommending the top 3 to fix next.
The payback calculation is the business case
For each top-3 item, the agent estimates effort to fix in developer-weeks against developer-weeks saved per quarter going forward. Fixing the data layer costs 40 dev-weeks but saves 20 dev-weeks per quarter, so it pays back in two quarters. That reframes the conversation. Instead of "PM says fix this," engineering sees a specific description, the affected systems, and a payback math they can weigh against feature work.
The report also includes a Debt Lifecycle section that tells you whether the team is creating new debt faster than it is fixing old debt. That is the signal that matters most long-term, and it is invisible without this kind of tracking.
How to use the weekly output
Treat the top three items as direct inputs to sprint planning. Engineering should be able to pick the top item and start immediately, because the agent outputs a specific description, the affected systems, and the business case. Pair it with the Auto Bugfix agent to handle the high-frequency bugs that surface once you clear the structural debt.
To run it, you need Jira or Linear with debt tickets tagged and linked to features, at least four past sprints with story points and actual completion time, the upcoming feature backlog mapped to systems, and engineering input on qualitative pain. Get those in place, schedule it for Monday at 10 AM, and let the first run tell you which debt item is really costing you velocity.