Adds an inner-dialogue step so the AI critiques its own draft and answers like a real person instead of a helpful assistant.
Builds production-grade Playwright workflows for scraping, form filling, screenshot/PDF capture, session reuse and anti-detection.
Analyzes marketing campaigns with five attribution models, funnel drop-off detection, and full ROI/ROAS/CPA metrics via three dependency-free Python CLIs.
A blunt Marc Andreessen persona that pressure-tests ideas and ventures market-first and ends with a BUILD / DERISK / KILL verdict plus confidence level.
Interviews a founder across 12 phases and materializes the whole company as a linked, version-controllable OKF markdown bundle.
A frontend discipline skill that forbids saying "done" until the site builds, passes checks, and is visually verified in a real browser.
Runs developmental, evidence-traceable reviews of papers, protocols and research ideas, plus local audits of low-stakes assessment rubrics.
A pharmacometrics toolkit covering NCA, compartmental fitting, popPK dataset QC, regimen simulation, exposure-response, bioequivalence, DDI and Bayesian TDM.
Plans, AI-generates, and assembles research talk decks — conference presentations, seminars, and thesis defenses — into PDF or PowerPoint.
Builds a manuscript-ready research packet — 60 verified, deduplicated references plus evidence matrix and BibTeX — using Parallel Search, Extract, and Research.
Builds clinical prediction pipelines (mortality, readmission, drug recommendation, sleep staging) on EHR and signal datasets using PyHealth's standard 5-stage workflow.
Builds and validates research-only CDS artifacts—GRADE evidence profiles, aggregate cohort tables, survival plans, model/biomarker evaluations, and de-identification checklists—with a hard no-patient-care boundary.
An all-in-one geospatial skill that turns satellite imagery, GIS, spatial statistics and spatial ML tasks into ready-to-run code patterns.
An Exa-powered skill for semantic web search and batch URL/PDF extraction, tuned for scientific and technical sources.
Design, build, and audit truthful, accessible, publication-ready scientific figures with Matplotlib, Seaborn, or Plotly.
Picks the right study design and generates seeded randomization schedules and DOE matrices before any data is collected.
Guides Claude through Hugging Face Transformers work: Hub model loading, pipeline inference, text generation, and Trainer fine-tuning on transformers 5.x.
Search OpenAlex, PubMed, and Google Scholar, extract accurate metadata, and generate, clean, and validate BibTeX bibliographies.
Queries live GenSpectrum LAPIS instances to report which pathogen lineages are circulating now, how fast they grow, and which mutations they carry — always with provenance.
A skill for scaling pandas/NumPy workloads beyond RAM using Dask's parallel and distributed collections.