shandley

Load data from files (CSV, JSON, JSONL, Parquet) or stdin for analysis and visualization with ggterm. Use when reading datasets, importing data, opening files, or when the user mentions loading, reading, or opening data.

Customize plot aesthetics using natural language. Use when the user wants to change colors, fonts, titles, labels, themes, or any visual aspect of a plot before publication.

Quick reference for ggterm capabilities. Use when the user asks what they can do, what commands are available, what plot types exist, how to use ggterm, or asks for help.

Search and retrieve plots from history. Use when the user asks about previous plots, wants to find a plot they made earlier, re-display a past visualization, or export a historical plot.

Generate markdown reports with embedded ggterm visualizations. Use when creating analysis reports, documenting results, exporting findings, or when the user wants plots in markdown format for sharing or documentation.

Create terminal data visualizations using Grammar of Graphics. Use when plotting data, creating charts, graphing, visualizing distributions, or when the user mentions plot, chart, graph, histogram, scatter, boxplot, or visualize.

Export terminal plots to publication-quality formats (PNG, SVG, PDF, HTML). Use when the user wants to save, export, publish, or create a high-quality version of a plot.

Apply publication-quality style presets to plots. Use when the user wants to style a plot like Wilke, Tufte, Nature, The Economist, or apply minimal/publication styling.

Audit, fix, and format bibliographies for academic manuscripts. Cross-references inline citations against the reference list to find orphans in both directions. Complements scientific-style (citation prose quality) and submission-prep (formatting checklists).

Generate and review scientific figures using AI. Creates publication-quality figures aligned with manuscript content using Gemini for generation and Claude vision for review. Works well with submission-prep and manuscript-writing skills.

Structure academic research papers using IMRaD format. Use when writing research manuscripts, organizing paper sections, drafting abstracts, or constructing titles. Works well with human-writing and scientific-style skills.

Respond to peer review comments and manage manuscript revisions. Use when drafting response letters, tracking changes across versions, or addressing reviewer concerns. Works well with scientific-style and human-writing skills.

Calibrate scientific claims and integrate citations properly. Use when adjusting claim strength, adding hedging language, integrating literature, or matching evidence to conclusions. Complements human-writing by addressing when hedging IS appropriate.

Prepare manuscripts for journal submission. Use when writing cover letters, checking submission requirements, or preparing figures and tables. Works well with human-writing and manuscript-writing skills.

Review a scientific figure across three dimensions: aesthetics (Wilke/Tufte), accuracy (claims match underlying data), and interpretability (main message is visually dominant). Use when the user asks to review, critique, audit, check, or polish a figure or figure script. Also use for "does this figure work?", "what's wrong with this plot?", or "is this publication-ready?".

Prime a session with the WashU HPC computing environment (HTCF and RIS Compute2) so cluster work can be managed without re-explaining the setup. Covers connection, filesystems and quotas, conda/LMOD software, SLURM partitions and sbatch headers, containers, job monitoring, and the failure modes specific to each cluster. Use when the user asks about submitting, monitoring, or debugging jobs on HTCF or RIS, installing software on the cluster, cluster storage, or SSH/VPN access to WashU compute.

Gracefully prepare a Claude Code session for a reboot or restart. Captures durable state before context is lost: updates project memories, records running HPC jobs, checks git status, flags time-sensitive timers, and prints a restart checklist with the first action to take after coming back. Use when the user says they are about to reboot, restart, shut down, or end a session, or asks to "save state before I restart".

Verify scientific identifiers (UniProt accessions, Pfam IDs, DOIs, PMIDs, NCBI accessions) in a project against authoritative sources. Use when generating reference data (positive controls, seed lists, citations), before committing identifier lists, or auditing an existing project for hallucinated accessions. Trigger when the user says "verify references", "check accessions", "audit my IDs", or asks whether protein/paper IDs are correct.

Diagnose microbiome/virome count data and recommend differential abundance analysis methods. Use when user wants to analyze their data, asks which method to use for their specific dataset, or provides count data files for analysis.

Guide for differential abundance analysis method selection, threshold choices, and result interpretation. Use when user asks about LinDA, ZINB, Hurdle, NB methods, q-value thresholds, effect sizes, sensitivity, FDR, or how to interpret DAA results. Also use when user runs an analysis and gets unexpected results (like 0 significant features).

Interpret differential abundance analysis results. Use when user has DAA output and wants to understand significant features, effect sizes, confidence levels, or next steps.

Write documents in a natural, human style. Use when writing essays, papers, documentation, reports, articles, or any prose content. Avoids common AI writing patterns.

Execute stenographic shorthand commands for efficient coding. Triggers on verb-colon-target patterns (dx, mk, ch, viz, stat, fork, switch, compare, merge, abandon) and steno-colon commands for session management. Branching commands (fork, switch, compare, merge, abandon) operate on .steno/graph.json, NOT git.

Research scientific papers from the last 30 days across OpenAlex, Semantic Scholar, PubMed, arXiv, and HuggingFace