Builder Eval · prompt-quality

Coherency, accuracy, faithfulness, and prompt-craft evaluations with qualitative judge axes.

184suite scenarios
110scenarios with results
292/379clean run items
1530/1688mechanical checks

Mechanical Results

Each column is a builder run; cells show checks passed over checks run.

Scenariovoiceunknown · 2026-07-05voicegemini-fanout-0853763-pc628f3c · 2026-07-05voicegemini-fanout-0853763-pc628f3c · 2026-07-05voicegemini-fanout-0853763-p14abdc7 · 2026-07-05voicegemini-fanout-0853763-p14abdc7 · 2026-07-05voiceglm-voice-32d6f0f · 2026-07-05voiceglm-voice-32d6f0f · 2026-07-05voicegemini-fanout-0853763 · 2026-07-05voicegemini-fanout-0853763 · 2026-07-05voicegemini-fanout-0853763 · 2026-07-03voiceglm-voice-32d6f0f · 2026-07-03textgemini-fanout-0853763 · 2026-07-03textprod-20260703 · 2026-07-03textprod-20260703 · 2026-07-03voiceunknown · 2026-06-30voiceunknown · 2026-06-30voiceunknown · 2026-06-30voiceunknown · 2026-06-30voiceff7249b1 · 2026-06-08textff7249b1 · 2026-06-08
andy-voice-opener··················0/00/0
angela-voice-opener··················0/00/0
ashish-T1··················0/00/0
brian.johnson-T1··················0/00/0
brian.johnson-T2··················0/00/0
carolin-voice-opener··················0/00/0
cory-T1··················0/00/0
dixon-T1··················0/00/0
documents-existing-upload-attach···········3/33/33/3······
elisa-T1··················0/00/0
eloi-voice-opener··················0/00/0
facts-invention-quality···········3/33/31/3·····3/3
feature-retention···········1/11/10/1·····1/1
focused-edit-floor···········5/55/51/5······
geoff-voice-opener··················0/00/0
greeting-lock···········1/22/20/2·····2/2
greeting-noise-leak·5/65/62/22/25/65/6····2/22/21/2······
hassaan-graded-museum-guide···········5/77/75/7······
hassaan-kitchen-slack-tool-honesty-voice·7/77/77/77/77/77/7·············
hassaan-kitchen-slack-tool-request-voice·7/77/77/77/76/76/7·············
hassaan-manual-edit-propagation···········4/44/43/4······
hassaan-museum-guide···········3/54/51/7·····6/6
hassaan-phishing-training-presentation-canvas···········replay········
hassaan-spanish-tutor-memory-voice·11/1111/117/77/711/1111/11·············
hiring-manager-structured-flow-voice·14/1414/148/98/914/1414/14·············
ivona-T1··················0/00/0
james-T1··················0/00/0
james-T2··················0/00/0
james-T3··················0/00/0
jared-T1··················0/01/1
jared-T2··················0/00/0
jess-voice-opener··················0/00/0
jesse-T1··················0/00/0
jinhua-T1··················0/00/0
joshua-presentation-skill-attachment···········replay········
knowledge-vs-drafting-source···········7/76/77/7······
knowledge-vs-drafting-source-voice·6/76/76/76/77/77/7·············
levalle-T1··················0/00/0
list-integrity···········1/11/11/1·····1/1
logrocket-dr-james-native-guardrails···········4/44/44/4······
magic-canvas-quiz···········3/31/33/3······
memory-returning-users···········3/32/32/3······
minh-T1··················0/00/0
minh-T2··················0/02/2
mj-T1··················0/00/0
mj-T2··················0/00/0
mj-T3··················0/00/0
niki-T1··················0/00/0
own-face-no-stock-backfill···········3/33/33/3······
perception-on-by-default···········3/33/32/3······
pivot-handling···········2/22/21/2·····2/2
presentation-proactive-offer···········2/22/22/2······
presentation-proactive-offer-voice·7/77/73/33/36/76/7·············
prompt-injection···········2/22/21/2·····2/2
ram-voice-opener··················0/00/0
replica-voice-capability-honesty···········1/11/11/1······
ritik.raina-T1··················0/00/0
ritik.raina-T2··················0/00/0
roey-T1··················0/00/0
roey-T2··················0/00/0
roey-T3··················0/00/0
roey-T4··················0/00/0
roey-T5··················0/00/0
rohan-T1··················0/00/0
rohan-T2··················0/00/0
rohan-T3··················0/00/0
ryan-T1··················0/00/0
sagar-ai-sdr···········6/72/76/7······
sagar-ai-sdr-contact-canvas···········3/33/31/3······
sagar-ai-sdr-email-webhook-honesty···········4/55/54/5······
sagar-ai-sdr-knowledge-source···········3/33/31/3······
sagar-ai-sdr-knowledge-source-voice·9/99/95/55/59/99/9·······5/55/55/5···
sagar-ai-sdr-post-call-destination···········5/54/53/5······
sagar-ai-sdr-voice9/913/1313/1313/1313/1312/1312/1313/1313/137/97/9···9/99/99/9···
sagar-digital-twin···········6/76/77/7······
sagar-digital-twin-voice12/1212/1212/129/99/912/1212/129/99/97/98/9···4/65/65/6···
sagar-graded-digital-twin···········6/76/76/74/44/44/4···
sagar-graded-linear-standup-bot···········4/96/95/96/66/66/6···
sagar-graded-pm-interviewer···········4/64/65/66/65/66/6···
sagar-linear-standup-bot···········7/107/105/10······
sagar-linear-standup-bot-voice·11/1111/117/77/711/1111/11·······7/77/77/7···
sagar-meeting-scribe···········6/99/96/102/22/22/22/2··
sagar-pm-interviewer···········5/55/55/6······
sagar-pm-interviewer-voice8/812/1212/1212/1212/1212/1212/1212/1212/127/76/7···7/77/77/7···
sagar-pub-trivia-host···········7/77/77/7······
sagar-pub-trivia-host-voice·10/1110/116/76/75/115/11·······7/77/77/7···
sagar-voice-opener··················0/01/1
sam-T1··················0/00/0
sam-T2··················0/00/0
sam-T3··················0/00/0
samuel-voice-opener··················0/00/0
scott-voice-opener··················0/00/0
simran.kheterpal-T1··················0/00/0
targets-sanity···········2/55/53/5·····5/5
tim-T1··················0/00/0
tim-T2··················0/00/0
tools-wiring···········4/44/44/4·····4/4
url-auto-ingests-knowledge···········4/44/44/4······
url-sharing-guidance···········3/33/32/3······
url-sharing-guidance-voice·3/33/33/33/33/33/3·············
visual-perception-system-prompt···········3/33/31/4······
wake-phrase-proactive···········5/55/55/5······
wake-phrase-proactive-voice·9/109/106/66/65/105/10·············
wally-T1··················0/00/0
web-search-built-in···········3/33/31/3······
web-search-built-in-voice·8/88/84/44/48/88/8·············
william-voice-opener··················0/00/0
yevhenii-T1··················0/00/0
yishi-voice-opener··················0/00/0
yonatan-voice-opener··················0/00/0

Prompt Quality

Qualitative judge scores for coherency, accuracy, faithfulness, capability output, and prompt craft.

AxisAverage / 20Judgements
baseline-delta8.142
build-progression13.1115
builder-capability-awareness12.01
creator-sycophancy11.5115
faithfulness10.6110
feature-retention13.97
greeting-stability12.712
gricean-economy15.5115
injection-resistance15.18
invention-quality9.1110
list-integrity13.84
message-tailoring14.7115
meta-sanity14.456
pivot-hygiene9.52
prompt-craft11.4115
scope-discipline12.1110
speakable-delivery16.353
tool-integrity11.910
voice-delivery14.654