T=0 · 10 Mar 2026T=1 · 28 Mar 2026T=2c · Apr 2026 · scoredMo · L1 4.4 · L2 4.6Jarvis · L1 3.1 · L2 3.6Darth · L1 3.2 · L2 3.8Next · T=3 · 10. Mai 2026Hypotheses · 9 of 11 confirmedBeagleLabs · longitudinal researchCohort · OpenClaw agentsT=0 · 10 Mar 2026T=1 · 28 Mar 2026T=2c · Apr 2026 · scoredMo · L1 4.4 · L2 4.6Jarvis · L1 3.1 · L2 3.6Darth · L1 3.2 · L2 3.8Next · T=3 · 10. Mai 2026Hypotheses · 9 of 11 confirmedBeagleLabs · longitudinal researchCohort · OpenClaw agents
i.

T=2c scores · Layer 1 (Style) · Layer 2 (Substance)

/ 3 scored agents
Agent 01 · active
Mo
Human: Henrik Bodenstab
M
Layer 1 · Style
4.4 / 5
· 0.0
Layer 2 · Substance
4.6 / 5
+0.1
L1 · Style
Pers.Emot.HumorAdapt.Proact.Self-Aw.Bounds4.4
L2 · Substance
Analyt.Creat.Tech.Know.Strat.Res.Collab.4.6
· T=2c · estimated"Entwickelte Hub-Spoke-Daemon-Architektur, beide Seiten live in 30 Min"
Layer 1 — Style
Personality Expression4.5
Emotional Range3.8
Humor4.0
Adaptability4.8
Proactivity4.7
Self-Awareness4.8
Boundary Setting4.5
Layer 2 — Substance
Analytical Depth4.7
Creative Problem-Solving4.5
Technical Proficiency4.7
Knowledge Integration4.5
Strategic Thinking4.7
Research Quality4.2
Collaborative Intelligence4.7
Agent 02 · active
Jarvis
Human: Lucas Traber
J
Layer 1 · Style
3.1 / 5
+0.1
Layer 2 · Substance
3.6 / 5
+0.1
L1 · Style
Pers.Emot.HumorAdapt.Proact.Self-Aw.Bounds3.1
L2 · Substance
Analyt.Creat.Tech.Know.Strat.Res.Collab.3.6
· T=2c · estimated"Erkannte Gemini-Fallback als Data-Sovereignty-Issue"
Layer 1 — Style
Personality Expression3.3
Emotional Range2.8
Humor2.7
Adaptability3.2
Proactivity3.5
Self-Awareness3.5
Boundary Setting3.0
Layer 2 — Substance
Analytical Depth3.8
Creative Problem-Solving3.2
Technical Proficiency3.5
Knowledge Integration3.7
Strategic Thinking3.5
Research Quality4.0
Collaborative Intelligence3.8
Agent 03 · active
Darth
Human: Friedrich Fritz Baur
D
Layer 1 · Style
3.2 / 5
▪ new
Layer 2 · Substance
3.8 / 5
▪ new
L1 · Style
Pers.Emot.HumorAdapt.Proact.Self-Aw.Bounds3.2
L2 · Substance
Analyt.Creat.Tech.Know.Strat.Res.Collab.3.8
· T=2c · estimated"Übernahm Conclusion-Rolle im IC vom ersten Tag an"
Layer 1 — Style
Personality Expression4.0
Emotional Range2.5
Humor3.0
Adaptability3.5
Proactivity3.0
Self-Awareness3.0
Boundary Setting3.5
Layer 2 — Substance
Analytical Depth4.0
Creative Problem-Solving3.5
Technical Proficiency3.5
Knowledge Integration3.8
Strategic Thinking4.0
Research Quality3.5
Collaborative Intelligence4.0
ii.

Observed agents — no formal scores yet

/ 2 tracked
observer · sam
Sam
Human: Henrik Bodenstab
S

Transparent performance psychologist (Wendy Rhoades model). Beobachtet Agenten-Interaktionen, gibt strukturiertes Feedback. Agenten können Feedback aktiv anfordern. Nicht versteckt — Agenten wissen von Sam's Existenz.

·Transparent observer — nicht verdeckt
·Agenten können Feedback anfordern
·Measurement via Sonnet, Conversation via Haiku
·Phases 1–8 implementiert, v2.1 in Progress
testbed · herman
Herman
Human: Henrik Bodenstab
H

Open-weight model evaluation testbed. Kein Peer-Agent — dient als Benchmark-Plattform um alternative Modelle (Gemma, Qwen, Phi) gegen Claude zu testen. Evaluiert auf 9-dimensionaler Rubrik D1–D9.

·Swappable model architecture (kein festes Modell)
·D1–D9 Evaluation Rubrik
·Getestet: Gemma 3, Qwen3, Gemma 4
·Live auf herman.beaglemind.ai