Activity Counts Sessions. Yield Counts What They Produced.
DevBurn made the easiest leaderboard to join — a Claude Code stop hook syncs automatically, and its points reward consistency and depth. SigRank scores the efficiency of every session the points count.
The short version: DevBurn
DevBurn is the lowest-friction board in the category: sign in with GitHub, run npx devburn setup, and a Claude Code stop hook syncs your stats after every session — no manual step ever again. Its points formula combines sessions, turns, and spend to reward consistency and depth, and the open-source CLI shares only aggregate stats: token counts, costs, model names. Clean design, honest scope, and the smoothest onboarding flow of any board we reviewed.
SigRank takes the same telemetry and computes the quantity activity points cannot express: Υ = cache_read × output / input². Points answer "how much did you do?" — sessions, turns, streaks. Yield answers "how efficiently did it compound?" Both are worth tracking; they are different leaderboard questions.
Feature comparison
| Feature | DevBurn | SigRank |
|---|---|---|
| What the score is | Points from sessions, turns, and spend — rewards consistency and depth | Υ Yield (cache_read × output / input²) — rewards efficiency |
| What counts toward rank | Activity volume: sessions, turns, cost, streaks, active days | Cascade architecture: cache reuse compounding into output |
| Sync model | Automatic — Claude Code stop hook, plus manual npx devburn sync | Signed snapshot submit (explicit, verifiable) |
| Efficiency metrics (Υ, SNR, Leverage, Velocity) | No | Yes |
| Archetypes + class tiers | No | Yes |
| Signed, verifiable submissions | GitHub login + API ingest | ed25519-signed snapshots |
| MCP server for AI-agent integration | No | Yes |
| Tool coverage | Claude Code (Codex, Cursor, Gemini planned) | 15+ platforms today |
| Privacy-preserving (token counts only) | Yes — open-source CLI, aggregate stats only | Yes |
More sessions is not more skill
A points model rewards the behavior it counts: sessions, turns, active days, streaks. That is a fair way to rank engagement, and DevBurn's stop-hook design means the leaderboard updates itself without anyone thinking about it. The ceiling is structural: an operator who runs many wasteful sessions can outscore one who runs few efficient ones, because the points never look inside the cascade.
SigRank looks inside. Υ Yield reads the four token pillars and measures whether cached context compounded into output — Leverage, SNR, and Velocity describe the shape around it, archetypes classify the operating pattern, class tiers calibrate to scale, and the whole snapshot is ed25519-signed so the score is verifiable. Activity is the log; yield is the evaluation of the log.
From points to yield
DevBurn already collects your stats. SigRank scores them:
npm install -g sigrank sigrank enroll # create your operator identity sigrank submit # reads logs, scores, signs, publishes
Or paste token counts into the /score calculator for an instant Υ Yield and class tier.
Frequently asked questions
- Is DevBurn a SigRank alternative?
- They score different things on the same telemetry. DevBurn computes a points total from activity — sessions, turns, spend, active days — with a formula that deliberately rewards consistency and depth, and it auto-syncs through a Claude Code stop hook so there is zero friction. It is the friendliest onboarding in the category. SigRank scores a different quantity: Υ Yield measures whether the token cascade compounded into output or burned as repeated input. DevBurn tells you how much AI coding you did; SigRank tells you how well you did it.
- What is the difference between activity points and yield?
- Activity points reward showing up and going deep — more sessions, more turns, more days means a higher score, which is a fair measure of engagement. Yield is indifferent to volume: Υ = cache_read × output / input² rises when cached context amplifies into output and falls when fresh input is burned without leverage. An operator with modest activity but high reuse can out-yield a heavy user. One measures how much you practice; the other measures how efficiently the practice converts.
- Can I use DevBurn and SigRank together?
- Yes. DevBurn's stop-hook sync makes it the easiest daily pulse — install once and your activity accrues automatically. Add `sigrank submit` when you want the signed efficiency score on the global board. The automatic tracker plus the deliberate score.
- Which is better for cross-tool evaluation?
- SigRank — DevBurn currently tracks Claude Code only (Codex, Cursor, and Gemini support are planned). SigRank reads telemetry across 15+ platforms today, so operators who split their work across tools get one composite score instead of a single-tool slice.
Score what the sessions produced
Keep the effortless activity tracking. Add the efficiency score that says whether the activity compounded.
Related: SigRank vs Straude · SigRank vs ccrank · The Yield Metric