performa

Engineering performance intelligence

See who is performing — and who is vibe coding.

Performa reads the work your team already ships — code, reviews and tickets — then scores outcomes first and effort second. Commit volume is not a ranking. AI dumps that look like delivery get a yellow fingerprint — and a vibe-coding flag when the physics of the work does not add up.

Manager decision support. Never an automated judgement.

Get started

Request access

If your company is not set up yet, we will put you on the waiting list and reach out.

Already have an account? Sign in.

Outcomes first

Performance is 1–100

Results stay raw credited delivery. Clean work without vibe coding reads as performing. Bug loops and AI dumps pull the score down.

AI usage

Yellow, on purpose

Impact-weighted share of merged work with a trailer, style score, or physics-scale change. High AI usage is not automatically vibe coding.

Vibe coding

Physics is definitive

Bursts and impossible velocity flag the work — a receipt is extra evidence, not a gate. The chip is violet so it never hides inside the yellow bar.

Interactive demo · Harbor · Aug 2026

The department a manager actually sees

Mocked Harbor data — same department map, individual scorecard and activity graph as the live product.

Demo data

Engineers analysed

7

7 of 7 ready · 1 team

Assessed work

100%

148 of 148 merged changes rated

Ticket linkage

91%

13 merged changes untraced

Delivering

3

High performance and high effort

Department map

Performance versus effort

Performance above 50 grinding quiet
Efficient
Delivering
Quiet
Grinding
LO Lena Ortiz · P 86 · R 412 / E 21
PS Priya Shah · P 81 · R 198 / E 9
AV Adele Voss · P 78 · R 261 / E 16
JR Jonah Reed · P 72 · R 305 / E 18
SO Sam Okonkwo · P 48 · R 220 / E 28
TK Theo Kane · P 38 · R 890 / E 24
MC Mira Chen · P 22 · R 14 / E 3
Lower effort Higher effort

Individual · Aug 2026

Theo Kane

Theo Kane · platform

Grinding

Scorecard · this period

94%

AI usage · this period

100%

Assessed · every merge rated

890

Results · this period

38

Performance · 1–100

24

Effort · this period

1.1h

Avg active hours · estimated time working

Bug loop

Clear

No Testing → In Progress returns

Vibe coding

Active

Confirmed bursts · −40% on Performance

Results & effort over time · 1 Aug to 31 Aug

Results 42.4 Effort 1.1 avg pts / working day 6 silent weekdays 4 short active days
0 pts 60 pts 120 pts 180 pts 4 Aug — short active day 5 Aug — silent weekday 10 Aug — silent weekday 13 Aug — short active day 14 Aug — silent weekday 19 Aug — silent weekday 21 Aug — short active day 26 Aug — silent weekday 28 Aug — short active day 3 Aug · Results 140 3 Aug · Effort 12 4 Aug · Results 95 4 Aug · Effort 8 6 Aug · Results 20 6 Aug · Effort 15 7 Aug · Results 160 7 Aug · Effort 10 11 Aug · Results 30 11 Aug · Effort 18 12 Aug · Results 180 12 Aug · Effort 9 13 Aug · Results 70 13 Aug · Effort 6 17 Aug · Results 15 17 Aug · Effort 12 18 Aug · Results 110 18 Aug · Effort 8 20 Aug · Results 25 20 Aug · Effort 20 21 Aug · Results 90 21 Aug · Effort 7 24 Aug · Results 10 24 Aug · Effort 8 25 Aug · Results 150 25 Aug · Effort 11 27 Aug · Results 18 27 Aug · Effort 14 28 Aug · Results 80 28 Aug · Effort 6 31 Aug · Results 12 31 Aug · Effort 10
1 Aug 8 Aug 15 Aug 22 Aug 31 Aug

Same-day points — bar height scales to this person's busiest day. Amber bands are silent weekdays, rose marks a short active day. Weekends are unmarked gaps.

Individual metrics

What each number means

The same six readings on every engineer. Volume is never the score.

Performance

1–100

From results and effort together. Clean work reads as performing. Vibe coding and bug loops pull the score down.

Results

Delivery

Credited work that shipped this period — sized by difficulty, not by how many commits landed.

Effort

Iteration

Reviews written, revisions, and days actually moving work. High effort with low results is grinding, not winning.

AI usage

Yellow

Share of shipped work with an AI fingerprint. A high number is a signal, not a verdict — and not the same as vibe coding.

Active hours

Time

Estimated time working from session gaps. A burst of output in a short day is how vibe coding usually shows up.

Vibe coding

Physics

Bursts and impossible velocity. The chip is violet so it stays distinct from the yellow AI bar.

Verdicts

This period

Yellow is AI usage · violet is vibe coding · sky is active hours

Lena Ortiz

Lena Ortiz · platform

Performance

86

Results

412

Effort

21

AI usage

12%

Active hours

4h 10m
Delivering

Priya Shah

Priya Shah · payments

Performance

81

Results

198

Effort

9

AI usage

18%

Active hours

5h 40m
Efficient

Adele Voss

Adele Voss · growth

Performance

78

Results

261

Effort

16

AI usage

55%

Active hours

3h 48m
Delivering

Jonah Reed

Jonah Reed · mobile

Performance

72

Results

305

Effort

18

AI usage

41%

Active hours

3h 22m
Delivering

Sam Okonkwo

Sam Okonkwo · checkout

Performance

48

Results

220

Effort

28

AI usage

33%

Active hours

6h 12m
Bug loop

Theo Kane

Theo Kane · platform

Performance

38

Results

890

Effort

24

AI usage

94%

Active hours

1h 05m
Vibe coding

Mira Chen

Mira Chen · data

Performance

22

Results

14

Effort

3

AI usage

8%

Active hours

45m
Quiet

How it works

Connect work. Get insights. Take action.

01

Ingest the systems you already have

Connect the code host and issue tracker you already use. Identity resolution pairs the same human across tools. No spreadsheets, no self-scores.

02

AI analyzes the work to provide insights

Outcomes, effort, AI usage and vibe coding are read from the work itself. The output is evidence — not a ranking, and not a verdict.

03

Humans turn insights into actions

A manager takes the insight into a 1:1. Alternative explanations stay on the page. Nothing publishes without sign-off.

Limited onboarding

Due to high demand, new companies join from the waiting list.

Request access with a work email. If your account is not set up yet, we will contact you shortly.

Request access