SIGNALAF
◈ SigRank vs clawdboard

One Board Counts Spend. The Other Counts Return.

clawdboard is a free community board for how much AI coding you do and how often. SigRank asks a narrower question: how much output did each fresh input token buy? The two sit side by side without conflict.

What clawdboard is, in its own words

clawdboard is a free, open-source leaderboard for AI coding agents. Its leaderboard page invites you to sign in with GitHub and run npx clawdboard auth, and ranks developers on a cost column beside tokens, active days, streak and a project column. Each fact below links to the page it came from and shows the date we read it.

Default ranking column
Estimated cost (API-equivalent), with tokens, active days, streak and a project column beside it (source, read 11 November 2026)
Time views
today, 7d, 30d, month, ytd and a custom range (source, read 11 November 2026)
Registered developers
96 (source, read 11 November 2026)
Tools it reads
The agent CLIs listed on its stats page, read from local logs (source, read 11 November 2026)
What leaves your machine
Aggregate token counts and a model name per session, per its methodology section (source, read 11 November 2026)
Licence
MIT (open-source repository morgen-so/clawdboard-oss) (source, read 11 November 2026)

What clawdboard does better

  • Breadth of tools. It reads the long list of agent tools on its stats page. SigRank's cascade is built around token pillars, so a tool is only useful to it if those four counts can be read out.
  • Dollar framing. If your question is "what would this have cost at API prices?", a cost column answers it directly. SigRank does not rank on price.
  • Habit and social loop. Streaks and a project column give people a reason to come back and a place to point at what they ship.
  • A larger crowd. On 11 November 2026 its stats page reported 96 registered developers, more than the public SigRank board had on the same date.

The methodological difference

Both tools start from the same four token counts: input, output, cache-read and cache-write. clawdboard prices them. SigRank takes ratios of them. The headline metric is Υ Yield = Leverage × Velocity, where Leverage is cache-read per fresh input token and Velocity is output per fresh input token (a ratio, not a speed). SNR, the output share of fresh traffic, sits alongside.

Worked example (hypothetical numbers, for illustration only)

Operator AOperator B
Fresh input1M8M
Output4M4M
Cache-read200M200M
Leverage200 / 1 = 200200 / 8 = 25
Velocity4 / 1 = 44 / 8 = 0.5
SNR4 / (4 + 1) = 0.804 / (4 + 8) = 0.33
Υ Yield80012.5

Same output, same cache-read, and B sends eight times the fresh input. B also moves more tokens in total, so on a board that sorts by volume or by a cost that tracks it, B edges ahead. On Yield, A is 64 times higher. A streak would not separate them at all if both ran every day.

How big is the SigRank field?

Small. On 11 November 2026 the 30-day board had 7 public operators, so read SigRank percentiles as a snapshot of a few people, not a population norm. The all-time board is current, and the definitions are in the methodology.

Use clawdboard when, use SigRank when

If you want toReach forBecause
See how much AI coding you do, at API prices, against other developersclawdboardCost is its main column and the page is built around it.
Keep a daily habit goingclawdboardStreaks and active days sit next to cost on its board.
Track a team on one pageclawdboardIt has a Teams tab next to the individual board.
Find out whether your reused context is turning into outputSigRankLeverage and Velocity are built from exactly that ratio.
Compare two operators who spend about the sameSigRankSpend ties; Yield separates them.
Hand a verifiable score to someone elseSigRankSubmissions are signed with your device key and re-scored server-side.

A workflow that uses both

  1. Leave clawdboard syncing in the background. Check it for spend and streak.
  2. Once a week, run a SigRank dry run and read Leverage and Velocity.
  3. If clawdboard cost climbs while Velocity falls, you are sending more fresh input for the same output. Look at what got re-sent instead of cached.
  4. If cost is flat and Velocity climbs, the work is getting denser. That is the case clawdboard cannot show you.

Adding SigRank if you already use clawdboard

Nothing to remove. SigRank reads local usage separately, so clawdboard keeps working as it does today. Create a connect code in Settings on signalaf.com, then:

npx sigrank                    # look at your local usage first
npx sigrank enroll             # bind this device with the connect code
npx sigrank submit --dry-run   # see the payload; nothing is sent
npx sigrank submit             # publish a signed snapshot

Want the numbers without an account? The /score calculator takes your four token counts and returns Υ Yield, Leverage, Velocity and SNR in the browser.

Questions people ask about clawdboard and SigRank

Is a cost ranking the same as an efficiency ranking?
No. A cost figure is tokens multiplied by a price per model, so it grows with how much you run. It does not say how much output that traffic produced. Yield is computed from token ratios instead, so a light user can outrank a heavy one.
Why can a cost-ranked board and a token-ranked board disagree?
Because price per token depends on the model. Heavy cache reuse on a cheaper model counts for a lot of tokens and not many dollars, so a cost ranking and a token ranking can order the same people differently. Neither says how much output that traffic produced; that is the gap Yield is built to close.
Is a streak a skill signal?
A streak counts days of use, so it rewards regularity. A short test session and an eight-hour session both keep it alive. Treat it as a habit tracker, not a measure of how well you drive; Yield is the measure of output per fresh input.
Does SigRank replace clawdboard?
No. They answer different questions and read the same kind of local logs. clawdboard tells you how much you ran and how consistently. SigRank tells you how much output you got per unit of fresh input you sent. Many people will want both.
Will my cost rank on clawdboard match my SigRank rank?
Not necessarily, and often not. SigRank's Yield is leverage times velocity, computed from token counts and not from price. Two operators with the same dollar spend can sit far apart on Yield, and a small spender can outrank a big one.
Is the SigRank field big enough to compare against clawdboard's?
No. The public SigRank board is small, so treat its percentiles as a small-sample reference rather than a population norm. See the dated figure in the body of this page and the live board for current numbers.

See your own Yield

Keep clawdboard for spend and streaks. Paste four token counts into the calculator to see Υ Yield, Leverage, Velocity and SNR, or read how to measure AI coding efficiency.

Sources and read dates

clawdboard facts link to its leaderboard, its stats page and its repository, all read on 11 November 2026. The SigRank operator count is from the 30-day board, read on 11 November 2026. clawdboard changes often; check its pages for anything that matters to you.

Related: Token Tracking Tools · Yield Calculator · SigRank vs Tokscale · The leaderboard · Methodology

What the July 2026 field showed

In a frozen field of 1,642 operators, the highest-Yield operator ranked #697 by volume, and 92.4% of the median operator’s tokens were cache reads. Volume and efficiency are different questions. The findings: Volume Isn’t Yield · The Operator Explains 95% of the Difference · The Three Modes Behind the Board.