EXP-014LAB.LOG
active⚗
Context Packet Builder
Can a small, typed context bundle make agent handoffs more predictable across longer tasks?
AgentsContext
// Lab notebook
Questions with prototypes attached.
Small, time-boxed investigations into agent workflows, evaluation, and software interfaces. The useful results graduate; the rest leave good notes.
EXP-011LAB.LOG
prototype⚗
Visual Diff Narrator
A compact workflow that turns interface screenshots and diffs into a structured review checklist.
VisionQA
EXP-009LAB.LOG
active⚗
Rubric Stress Tester
Generating adversarial task variants to find ambiguous or easily gamed evaluation criteria before a benchmark ships.
EvalsBenchmarks
EXP-003LAB.LOG
archived⚗
Daily Research Digest
An early information-filtering experiment that taught me to value traceability over volume.
AutomationResearch