sciclaw

Generate a standalone image, a visual asset for another artifact, or a complete image-rendered presentation slide. Use when the user asks to generate, illustrate, visualize, make an image for a slide, improve a presentation with generated imagery, or create an image-generated slide. Preserve the explicitly requested output format.

Ship a SciClaw software change through an isolated worktree, implementation, pruning, verification, commit, push, draft pull request, CI closeout, and independent review. Use when asked to ship a job, complete a change through PR, prepare a release candidate, or follow the repository's happy path toward Homebrew release and Data3 deployment.

Use this skill whenever scientific image assets need deterministic preprocessing (resize, crop, convert, DPI normalization, montage/contact sheets) using ImageMagick 7 `magick`.

Use this skill whenever `.pptx` slide decks are created, edited, summarized, templated, merged, or analyzed.

Use this skill whenever spreadsheet files (`.xlsx`, `.xlsm`, `.csv`, `.tsv`) are created, edited, cleaned, analyzed, or converted.

Read, edit, and diff Word documents (.docx) with tracked changes and comments using the docx-review CLI — a .NET 8 tool built on Microsoft's Open XML SDK. Ships as a single 12MB native binary (no runtime). Use when: (1) Adding tracked changes (replace, delete, insert) to a .docx, (2) Adding anchored comments to a .docx, (3) Reading/extracting text, tracked changes, comments, and metadata from a .docx, (4) Diffing two .docx files semantically, (5) Responding to peer reviewer comments with tracked revisions, (6) Proofreading or revising manuscripts with reviewable output, (7) Any task requiring valid tracked-change .docx output with proper w:del/w:ins markup that renders natively in Word.

Inspect, schema, and fill true AcroForm PDFs using sciClaw's pdf_form_inspect, pdf_form_schema, and pdf_form_fill tools. Use when: (1) checking whether a PDF is a real fillable AcroForm, (2) listing form field names, types, and choices, (3) filling a supported form from structured values, (4) preparing reviewable insurance, legal, administrative, or clinical form drafts, (5) reporting skipped fields or unused input keys after a fill. Do not use for scanned PDFs, OCR tasks, or XFA-only forms.

Create "How X Works" deep-dive explainer websites with the curious, technical-but-accessible style of howkeyboardswork.com. Use when asked to explain how something works, create an educational single-page guide, or build an interactive explainer site on any topic.

Create clean first-draft Word documents from Markdown using pandoc, with sciClaw auto-applying the bundled NIH reference template unless a custom --reference-doc is provided.

De-identify clinical text for drafts, notes, and manuscript excerpts using phi-cleaner before sharing with external tools or collaborators. Use when: (1) Removing patient names/dates/identifiers from free-text notes, (2) Producing sanitized excerpts for LLM prompts, (3) Running a PHI detection pass to audit sensitive text, (4) Preparing de-identified appendices or examples for manuscripts and presentations.

Search PubMed, fetch article metadata, traverse citation graphs, and look up MeSH terms from the command line. Use when: (1) Searching PubMed with Boolean/MeSH queries, (2) Fetching article details by PMID (abstract, authors, DOI, MeSH terms), (3) Finding papers that cite a given article (cited-by), (4) Finding papers cited by a given article (references), (5) Finding related articles with relevance scores, (6) Looking up MeSH vocabulary (tree numbers, scope notes), (7) Exporting citations in RIS format for Zotero/EndNote, (8) Building reproducible literature review workflows.

Use this skill whenever PDF files are created, read, merged, split, transformed, OCR-processed, or data-extracted.

Rewrite AI-generated text to sound naturally human-written. Use when asked to humanize text, make writing sound less AI, bypass AI detection, rewrite for natural tone, or convert robotic/formal text to conversational style. Triggers on "humanize this", "make this sound human", "rewrite naturally", "anti-AI detection", or any request to remove AI tells from writing.

Define benchmark runs and log outcomes with consistent metrics, acceptance criteria, and reproducible artifact references.

Capture experiment provenance with reproducible run metadata, artifact pointers, and decision logs for scientific claims.

Build publication-grade Mermaid diagrams with consistent styling and SVG-first export for manuscript assets.

Search and summarize preprints from bioRxiv/medRxiv with explicit date windows and traceable query logs.

Author and render Quarto manuscript documents (.qmd) with consistent formatting and loop-by-loop updates.

Draft and revise scientific manuscript sections with claim-evidence alignment and reproducible method reporting.