For decades, LOC was a joke metric. You could game it. Verbose code looked more “productive” than elegant solutions. Senior devs knew better.
AI changes the equation.
When you're orchestrating agents across 100,000+ lines a week, LOC stops measuring your typing speed. It starts measuring your leverage — how much of a codebase your judgment, taste, and architectural intuition can actually reach.
Subplot tracks that leverage.
The old critique of LOC assumed you were writing the code. In that world, measuring lines rewarded the wrong behavior.
But when agents are writing the code, you're not the bricklayer anymore. You're coordinating crews. LOC touched isn't measuring your craftsmanship — it's measuring your span of control.
The question isn't “how productive is this developer?”
It's “how much surface area can one person's judgment effectively reshape?”
That's a number worth tracking.
Here's what separates Subplot from a line counter:
Your greenfield velocity is fast. A fresh 5K-line project? You're shipping 800 lines a day, easy. But what happens at 100K lines?
Most developers hit a wall. Architectural debt. Complex dependencies. Context that doesn't fit in anyone's head — human or AI. Velocity drops to 200 lines/day or worse.
Velocity retention measures what percentage of your speed you keep as the codebase grows.
Baseline velocity (first 60 days): 580 lines/day
Current velocity (last 30 days): 420 lines/day
Retention: 72%That 72% is the signal. It says your architecture stays promptable — clean enough that AI can still understand and extend it. Modules are coherent. Patterns are consistent. Context is localized.
The industry median for projects this size? 54%.
Velocity retention is a proxy for “can AI still help here?” — and that's the architectural question that actually matters now.
What would this have cost to build in 1981? Useful for investor conversations and board decks. We label it honestly: it's a historical reference point, not a prediction.
“I produced what traditional estimates would price at $2.1M.”
What would this realistically cost to hire contractors today? Modern calibration: 50 lines/hour, adjusted for complexity. When traditional feels inflated, this is your grounded comparison.
Among developers actively using AI tools, where do you stand? Not compared to 1981. Not compared to contractors. Compared to the current cohort of people figuring out how to operate at this new scale.
As the community grows, this becomes the benchmark: what does AI-assisted development actually look like?
The constraint has moved.
For decades, the bottleneck in software was writing code. Now agents can produce 1000x the output. The new constraint is your ability to direct and absorb that output — to maintain architectural coherence while operating at scale.
If you ran a small farm and suddenly had access to combines and 1000 acres, the goal is still to grow crops and make money. But the immediate question is: can you learn to operate at this scale?
The capability gap between those who figure it out and those who don't will be massive. And it's not optional — this capacity is available to everyone. Your competitors have the same access to the same tools.
For engineers whose goal is staying employed, the ability to operate at this scale is the bottleneck.
Subplot measures whether you're making the transition.
One view across all your repositories. Portfolio-wide metrics and individual breakdowns.
The number that reveals architectural quality without explicitly measuring it. Watch how your speed changes as projects mature.
Automatic classification: Code, Tests, Docs, Config. See exactly what you're building.
Full CLI for automation. Menu bar for quick stats without context switching.
Everything stays on your machine. SQLite database, no cloud sync, no telemetry. Your metrics are yours.
Opt-in only. Anonymous. No filenames, no commit messages, no code. Just aggregate metrics that power the Subplot Index.
Point Subplot at your git repositories.
Daily metrics via scc and git. Lines by language, commit activity, complexity.
Dashboard shows output against baselines, velocity retention over time, and your position in the Index.
That's it. No configuration. No tagging commits. Just visibility into what you're actually building.
Version 1.0 · macOS 14.0+
Download for MacFree to use · Direct download · No App Store required
Requirements: