Julia

Automated pipeline that extracts respiration traces (and calculates respiratory rate) from an iPhone video, stress-testing cheap hardware for measuring clinically relevant sleep parameters.

An orchestration layer that turns a research question into a structured, citation-anchored evidence digest. A Claude skill converts the question into a run scaffold and instruction file, Claude Code executes the run against Paperclip (an agent-first CLI over 11M+ full-text papers, trials and regulatory documents), and the result is a filled digest: evidence tables, synthesis, confidence reads and line-level verification anchors. The contribution is the structure rather than new software; the repo carries three complete example runs.

An observability layer for agent runs: a repository that turns opaque agent logs into deterministic readouts — input/output tables, color-coded views and a provenance report — so scientists can compare what an agent claims it did with what append-only logs show actually happened. Demonstrated on macrophage differentiation protocols extracted from papers, where three prompting approaches looked identical at eight papers but diverged sharply at 200: every run claimed full coverage, and only one achieved it. What you can see, you can begin to trust.