md-docs
This skill should be used ONLY when the user asks to update or initialize README.md, CLAUDE.md, or AGENTS.md. Trigger phrases include "update README", "init README", "update context files", "update CLAUDE.md/AGENTS.md". Do NOT activate for any other Markdown file updates.
What this skill does
# Markdown Documentation Management
## Overview
Manage project documentation for Claude Code workflows including README.md and agent context files (AGENTS.md / CLAUDE.md). This skill enforces a strict audience split: **README.md is for humans**, **AGENTS.md is for agents and developers running commands**. Use this skill when initializing new projects, updating existing documentation, or ensuring context files accurately reflect current code.
By default the skill operates **recursively** across the whole repository, not just the root. `update-*` workflows find and refresh every existing README.md / AGENTS.md in the tree; `init-*` workflows create files at the repo root and at each package root (a directory holding a manifest or workspace member) that lacks one. Each file is scoped to its own directory subtree and the nearest enclosing manifest. Pass `--root-only` to restrict any workflow to the repository root, or pass a `path` argument to limit the sweep to one subtree.
The skill emphasizes verification and validation over blind generation — analyze the actual codebase structure, file contents, and patterns before creating or updating documentation. All generated content should be terse, imperative, and expert-to-expert rather than verbose or tutorial-style.
## Audience Split
This is the central rule for everything this skill produces.
| File | Audience | Contains | Excludes |
| ------------------------------------- | -------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------- |
| `README.md` | Humans browsing the repo (GitHub, npm, etc.) | Description, badges, links to docs/site/demo, references, papers, related work, acknowledgments, license, contributing pointer | Any CLI commands, `just` recipes, package scripts, build/test/lint workflows, project structure trees, API reference |
| `AGENTS.md` (and `CLAUDE.md` symlink) | AI agents and developers working in the repo | Stack, commands (install, dev, build, test, lint), `just` recipes, `package.json` scripts, `Makefile` targets, code style, architecture, conventions, contribution workflow | Marketing copy, badges, external links unrelated to development |
When in doubt, ask: *would a human reading this on GitHub care, or only a developer/agent running commands?* If the latter, it goes in AGENTS.md.
## Workflow Selection
Pick the workflow that matches the user's intent:
| Trigger | Workflow | Reference |
| --------------------------------------------------------------------- | -------------------- | ----------------------------- |
| "update README" / "refresh README" (file already exists) | Update README | `references/update-readme.md` |
| "init README" / "create README" / "new README" (no file or `--force`) | Initialize README | `references/init-readme.md` |
| "update CLAUDE.md" / "update AGENTS.md" / "update context files" | Update Context Files | `references/update-agents.md` |
| "init AGENTS.md" / "create CLAUDE.md" / "init context" | Initialize Context | `references/init-agents.md` |
Selection rules:
- If the target file already exists and the user says "update" / "refresh" / "fix", route to an `update-*` workflow.
- If the target file is missing or the user says "create" / "init" / "new", route to an `init-*` workflow.
- For ambiguous requests, enumerate the target files first (see Recursive Discovery) and confirm with the user.
- If the user invokes the skill with no arguments, default to listing the files present across the tree and proposing a workflow rather than guessing.
- Multiple workflows in one request (e.g. "update README and AGENTS.md") are fine — run them sequentially in the order the user listed them, reporting each result independently.
- Each selected workflow applies to every target file the discovery step finds, unless `--root-only` or a narrowing `path` argument restricts the sweep.
## CONTRIBUTING.md Policy
This skill does **not** maintain `CONTRIBUTING.md`. If the workflow detects a `CONTRIBUTING.md` alongside any target file (repo root or any package root):
1. Do not block other writes; surface the advisory for that directory.
2. Recommend the user merge its contents into the sibling `AGENTS.md` (since AGENTS.md now owns the development workflow, branch conventions, review process, and tooling references).
3. Suggest deleting `CONTRIBUTING.md` after the merge so the agent context file is the single source of truth.
4. Do not auto-merge or auto-delete; the user performs the merge.
Report the recommendation per directory in the standard summary format and continue with whichever README/AGENTS workflow the user requested, ignoring the `CONTRIBUTING.md` file itself.
## Prerequisites
Before using any documentation workflow, verify basic project structure:
```bash
git rev-parse --git-dir
```
Ensure the output confirms you are in a git repository. If not initialized, documentation workflows may still proceed but git-specific features will be skipped.
Resolve the repository root once; all discovery is relative to it (or to a `path` argument, if given):
```bash
git rev-parse --show-toplevel
```
Then enumerate target files (see Recursive Discovery below) before attempting any workflow. If `CONTRIBUTING.md` shows up next to a target, apply the policy above for that directory.
## Recursive Discovery
By default, workflows act on every relevant file in the tree. `--root-only` collapses this to the repo root; a `path` argument scopes it to one subtree.
**Exclusions (always).** Skip these everywhere during discovery and creation:
- VCS and dependency/build output dirs: `.git`, `node_modules`, `vendor`, `.venv`, `target`, `dist`, `build`, `out`, `.next`, `coverage`.
- Anything ignored by git (rely on `--exclude-standard` / `git check-ignore`).
- Hidden dot-directories (`.github`, `.vscode`, `.claude`, …) — **unless** the directory contains a manifest.
**`update-*` discovery** — find existing files to refresh:
```bash
# README.md / AGENTS.md tracked or untracked, respecting .gitignore
git ls-files --cached --others --exclude-standard -- '**/README.md' 'README.md' '**/AGENTS.md' 'AGENTS.md'
```
Drop any path under an excluded dir. `CLAUDE.md` is a symlink to its sibling `AGENTS.md` and is never processed on its own.
**`init-*` discovery** — find package roots that should get a new file:
```bash
# Directories holding a language/tooling manifest = package roots
git ls-files --cached --others --exclude-standard \
-- '**/package.json' 'package.json' '**/Cargo.toml' 'Cargo.toml' \
'**/pyproject.toml' 'pyproject.toml' '**/setup.py' 'setup.py' \
'**/go.mod' 'go.mod' '**/foundry.toml' 'foundry.toml' \
'**/Gemfile' 'Gemfile' '**/composer.json' 'composer.json'
```
The set of package roots is the repo root plus the unique directories of those manifests (minus exclusions). `init-*` creates the target file only in package roots that lack it; it does not create files in arbitrary leaf directories.
**Per-file scoping.** Treat each target independently:
- Metadata source is the nearest enclosing manifest (the one in its own directory, else walk up to the repo root).
- A nested `READMERelated in Writing & Docs
jax-development
IncludedUse this skill when the user is writing, debugging, profiling, refactoring, reviewing, benchmarking, parallelising, exporting, or explaining JAX code, or when they mention JAX, jax.numpy, jit, grad, value_and_grad, vmap, scan, lax, random keys, pytrees, jax.Array, sharding, Mesh, PartitionSpec, NamedSharding, pmap, shard_map, Pallas, XLA, StableHLO, checkify, profiler, or the JAX repo. It helps turn NumPy or PyTorch-style code into pure functional JAX, fix tracer/control-flow/shape/PRNG bugs, remove recompiles and host-device syncs, choose transforms and sharding strategies, inspect jaxpr/lowering/IR, and benchmark compiled code correctly.
nature-article-writer
IncludedDrafts, rewrites, diagnostically critiques, and style-calibrates primary research manuscripts for Nature and Nature Portfolio journals. Use when the user wants a Nature-style title, summary paragraph or abstract, introduction, results, discussion, methods, figure legends, presubmission enquiry, cover letter, reviewer response, or when a scientific draft sounds generic, jargon-heavy, structurally weak, or AI-ish and needs precise, broad-reader-friendly prose without inventing data, analyses, or references. Best for primary research articles and letters rather than reviews or press releases unless explicitly adapting one.
deckrd
IncludedDocument-driven framework that derives requirements, specifications, implementation plans, and executable tasks from goals through structured AI dialogue. Use when user says "write requirements", "create spec", "plan implementation", "derive tasks", "structure this feature", "break down into tasks", or "document this module". Also use for reverse engineering existing code into docs (/deckrd rev). Do NOT use for direct code writing — use /deckrd-coder after tasks are generated. Do NOT use when the user only wants to run or fix existing code without planning.
clinical-decision-support
IncludedGenerate professional clinical decision support (CDS) documents for pharmaceutical and clinical research settings, including patient cohort analyses (biomarker-stratified with outcomes) and treatment recommendation reports (evidence-based guidelines with decision algorithms). Supports GRADE evidence grading, statistical analysis (hazard ratios, survival curves, waterfall plots), biomarker integration, and regulatory compliance. Outputs publication-ready LaTeX/PDF format optimized for drug development, clinical research, and evidence synthesis.
handling-sf-data
IncludedSalesforce data operations with 130-point scoring. Use this skill to create, update, delete, bulk import/export, generate test data, and clean up org records using sf CLI and anonymous Apex. TRIGGER when: user creates test data, performs bulk import/export, uses sf data CLI commands, needs data factory patterns for Apex tests, or needs to seed/clean records in a Salesforce org. DO NOT TRIGGER when: SOQL query writing only (use querying-soql), Apex test execution (use running-apex-tests), or metadata deployment (use deploying-metadata).
accelint-ac-to-playwright
IncludedConvert and validate acceptance criteria for Playwright test automation. Use when user asks to (1) review/evaluate/check if AC are ready for automation, (2) assess if AC can be converted as-is, (3) validate AC quality for Playwright, (4) turn AC into tests, (5) generate tests from acceptance criteria, (6) convert .md bullets or .feature Gherkin files to Playwright specs, (7) create test automation from requirements. Handles both bullet-style markdown and Gherkin syntax with JSON test plan generation and validation.