Shelves Context & AI engineering
Writing skills
1,028 rows from 545 repositories
Skills, slash commands and hooks: the artefacts that configure an agent.
What Writing skills skills agree on
388 skill files read, by 408 of the 422 authors on this shelf whose files we hold, 2026-09-06
The middle one of the 64 measured here is ~1.7k tokens long, counted with cl100k_base
Counted by distinct author, so one author publishing three of these counts once. Where a claim sits in fewer files than authors, that is said: a claim held by forty authors across three files is one file people copied, not forty people who agreed. Near-identical wordings are grouped and the other wordings are shown, so the grouping is yours to check.
What they tell the agent to do
- Keep SKILL.md under 500 lines118 of 408 in 104 filesalso worded as Keep the SKILL.md body under 500 lines; Keep SKILL.md body under 500 lines
- Write skill instructions in imperative form44 of 408 in 37 filesalso worded as Write instructions in imperative form; Write SKILL.md in imperative verb-first form
- Move detailed reference material into references files41 of 408 in 34 filesalso worded as move detailed reference material into separate files; Move detailed reference material into reference files
- Run init_skill.py to initialize new skills40 of 408 in 33 filesalso worded as Run init_skill.py to scaffold new skills; Initialize new skills with init_skill.py
- Draft assertions while test runs are in progress32 of 408 in 25 filesalso worded as Draft assertions while runs are in progress; Draft assertions while the runs execute
- Keep description under 1024 characters31 of 408 in 18 filesalso worded as Keep the description under 1024 characters; Keep descriptions under 1024 characters
- Run package_skill.py once the skill is complete29 of 408 in 23 filesalso worded as package the finished skill with package_skill.py; Run package_skill.py to validate and package
- Spawn with-skill and baseline runs in the same turn28 of 408 in 21 filesalso worded as Spawn with-skill and baseline runs in one turn; Spawn with-skill and baseline runs in same turn
- Test added scripts by actually running them27 of 408 in 21 filesalso worded as Test added scripts by running them; test scripts by actually running them
- Follow the creation steps in order24 of 408 in 17 filesalso worded as Follow the creation process steps in order; Follow the six creation steps in order
- Generate the eval viewer before evaluating outputs yourself24 of 408 in 17 filesalso worded as Launch the eval viewer before reviewing outputs yourself; Generate the eval viewer before reviewing outputs yourself
- Write the description in third person24 of 408 in 18 filesalso worded as Write descriptions in third person; write description in third person under 1024 characters
- Gather concrete usage examples before writing22 of 408 in 21 filesalso worded as gather concrete usage examples before building; Gather concrete usage examples from the user
- Capture the user's intent before drafting the skill21 of 408 in 14 filesalso worded as Capture user intent before drafting the skill; Capture the user's intent before drafting
- Keep references one level deep20 of 408 in 15 filesalso worded as Keep references one level deep from SKILL.md; Keep references one level deep and independently usable
What they tell it not to do
- Do not duplicate content between SKILL.md and references46 of 408 in 39 filesalso worded as Do not duplicate information between SKILL.md and references; Don't duplicate information between SKILL.md and references
- Do not add frontmatter fields beyond name and description34 of 408 in 21 filesalso worded as Do not add extra frontmatter fields; Do not add extra YAML frontmatter fields
- Do not write vague descriptions30 of 408 in 19 filesalso worded as Writing vague descriptions without triggers; Do not write vague instructions
- Do not write custom HTML for the viewer26 of 408 in 19 filesalso worded as Don't write custom HTML for the viewer; Don't write custom HTML for the review viewer
- Do not nest references deeper than one level23 of 408 in 17 filesalso worded as Do not nest references more than one level deep; Avoid deeply nested references
- Avoid asking too many questions in one message21 of 408 in 16 filesalso worded as Do not ask too many questions in one message; Don't ask too many questions in one message
- Do not create extraneous documentation files20 of 408 in 14 filesalso worded as Do not create extraneous documentation or auxiliary files; Don't create extraneous documentation files like README or CHANGELOG
- Do not create workspace directories upfront19 of 408 in 13 filesalso worded as Don't create all workspace directories upfront; Don't create workspace directories upfront
- Do not create misleading or malicious skills19 of 408 in 12 filesalso worded as Don't create misleading or malicious skills; Don't create misleading skills or enable unauthorized access
- Do not put when-to-use sections in the body18 of 408also worded as Don't put when-to-use info in the skill body; Do not put when-to-use info in the skill body
What they expect to be installed
- init_skill.py40 of 408 in 34 files
- package_skill.py39 of 408 in 34 files
- subagents32 of 408 in 24 files
- bash32 of 408 in 25 files
- jq28 of 408 in 21 files
- git27 of 408 in 25 files
- generate_review.py26 of 408 in 19 files
- python324 of 408
- grep22 of 408 in 21 files
- mkdir22 of 408 in 15 files
What they ask it to produce
- SKILL.md108 of 408 in 88 filesalso worded as SKILL.md file; SKILL.md draft
- Create packaged .skill file56 of 408 in 44 filesalso worded as .skill package; .skill package file
- SKILL.md with YAML frontmatter55 of 408 in 47 filesalso worded as SKILL.md with name and description frontmatter; SKILL.md with frontmatter
- eval_metadata.json per test case35 of 408 in 28 filesalso worded as eval_metadata.json; evals.json test cases
- benchmark.json and benchmark.md34 of 408 in 27 filesalso worded as benchmark.json; benchmark.json benchmark summary
- timing.json per run33 of 408 in 26 filesalso worded as timing.json; timing.json with run statistics
- grading.json per run33 of 408 in 26 filesalso worded as grading.json; grading.json per eval
- evals.json with test prompts26 of 408 in 20 filesalso worded as evals/evals.json; evals.json
When Writing skills authors say to reach for one
The situations these authors wrote into their own files, counted out of the same 408 authors, with the skills that name each one
- User wants to create a new skill134 of 408 in 118 files
- Skill creator
- Write a skill
- Pm skill creator
- Skill creator openai
- Skill development
- Agents skill creator
- and 197 more on this shelf
- User wants to run evals or benchmark a skill35 of 408 in 28 files
- User wants to optimize a skill's description for triggering35 of 408 in 28 files
- When creating a new skill27 of 408 in 26 files
- Skill creator
- Crypto skill creator
- and 3 more on this shelf
- User asks to create a hook26 of 408 in 20 files
- User asks to turn the current workflow into a skill25 of 408 in 18 files
- User says create, add, or build a skill24 of 408
- Write a skill
- Skill write skill
- Skill builder
- and 19 more on this shelf
- User wants to improve or refactor existing skills18 of 408
How Writing skills skills are built
943 skill directories by 517 authors, read from their repositories’ own file trees 2026-08-05
The middle bundle among those shipping files is 6 files, 37.1 KB beside SKILL.md
Counted by distinct author, same as above, so one author publishing forty template copies counts once. SKILL.md itself is not counted as a file, so a single-file skill is one where that file is the whole skill.
The shape
- SKILL.md is the whole skill144 of 517 authors, 263 of 943 skills
- files ship beside it373 of 517 authors, 680 of 943 skills
- executable scripts ship inside253 of 517 authors, 402 of 943 skills
The folders they converge on
- references/275 of 517 authors, 434 of 943 skills
- scripts/248 of 517 authors, 393 of 943 skills
- assets/142 of 517 authors, 186 of 943 skills
- agents/141 of 517 authors, 188 of 943 skills
- eval-viewer/108 of 517 authors, 138 of 943 skills
- examples/25 of 517 authors, 33 of 943 skills
anthropics/skills/skills/skill-creator Skill
in 1 stackno license168,934★ repoPublic repository for Agent Skills
yaojingang/yao-meta-skill Skill
2,355★YAO = Yielding AI Outcomes. A rigorous engineering, evaluation, governance, and portability system for reusable agent skills.
AIPexStudio/AIPex/packages/browser-runtime/src/skill/built-in/skill-creator-browser Skill
1,235★ repoAIPex: AI browser automation assistant, no migration and privacy first. Alternative to Manus Browser Operator、 Claude Chrome and Agent Browser
marketingjuliancongdanh79-pixel/skill-generator Skill
no license341★Bộ công cụ tạo AI Skill từ ý tưởng — Dành cho Antigravity, Claude Code, Cursor, Windsurf. Đánh giá 100/100 S-tier.
jaechang-hits/SciAgent-Skills/.claude/skills/sciagent-skill-creator Skill
no license320★ repo197 bioinformatics & life science skills for Claude Code and AI agents — BixBench 92.0% accuracy. RNA-seq, single-cell, drug discovery, proteomics, and more. Powers OmicsHorizon.
w95/awesome-claude-corporate-skills/00-meta/skill-creator Skill
150★ repo166 production-ready Claude AI skills organized by corporate role — executive leadership, finance, HR, marketing, sales, legal, operations, engineering, product, data, customer success, procurement & document processing
robertguss/claude-code-toolkit/.claude/skills/skill-creator Skill
103★ repoA collection of custom skills that extend Claude's capabilities with specialized workflows, methods, and domain knowledge.
cxcscmu/SkillLearnBench/baselines/b4-skill-creator/inject/skill-creator Skill
75★ repo[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
beriberikix/zephyr-agent-skills/.agent/skills/skill-creator Skill
59★ repoA complete catalog of Agent Skills (agentskills.io) for Zephyr RTOS development.
panaversity/agentfactory-business-plugins/.claude/skills/skill-creator Skill
28★ repoMarketplace of domain-specific plugins for AI agents (Cowork, Claude Code, OpenClaw). Build autonomous business workflows for finance, banking, legal operations, and sales using modular agent skills and commands.
a-tokyo/agent-skills/.agents/skills/skill-creator Skill
15★ repo🧠 AI Agent skills for LLMs and AI Agents - Claude, Codex, Cursor etc.
isomoes/skills/.agents/skills/skill-creator Skill
10★ repoCustom skills for agents to support academic paper writing and thesis management
s-hiraoku/synapse-a2a/.agents/skills/anthropic-skill-creator Skill
no license10★ reporohit-yadav34/second-mind/.agents/skills/skill-creator Skill
no license8★ repoSecond Mind is a skill that scans your repository and conversation history to capture design decisions, open questions, and session context, storing everything as plain Markdown so project knowledge is portable, human-readable, and easy to resume.
gmickel/better-skill-builder Skill
8★Skill that teaches AI agents how to build Agent Skills for Claude Code, Codex, Amp, and OpenCode. Production patterns from gno, sheets-cli, raindrop, outlookctl, and flow.
Victory-7291/project-scaffold-setup-skills/.agents/skills/skill-creator Skill
7★ repoReusable agent skills for scaffolding C++ and embedded projects
Poorgramer-Zack/dart-expert-skills/.agents/skills/skill-creator Skill
no license7★ repoA comprehensive library of modular Agent Skills for Flutter & Dart development
sofumel/claude-handoff-revive/plugins/handoff-revive/skills/handoff-revive Skill
6★ repoResume Claude Code work after rate/usage/context limits without replaying the prior transcript. Auto-saves at 90%/95% usage. Plugin-installable, 10 languages.
crowdin/skills/.agents/skills/skill-creator Skill
5★ repoAgent Skills to help developers using AI agents with Crowdin
yaniv-golan/skill-creator-plus/skill-creator-plus/skills/skill-creator-plus Skill
4★ repoBased on Anthropic's skill-creator — with bug fixes, working Cowork support, and official best practices baked in. Claude skill for creating, testing, and improving other Claude skills.
toolbox-playground/super-claudio/skills/ai-and-agents Skill
4★ repoA curated Claude Code plugin with skills for anything Claude can do.
KimYx0207/meta-skill-creator/skills/meta-skill-creator Skill
4★ repoA portable Agent Skill for designing, refactoring, and reviewing reusable skills.
xonika9/agent-skills/skills/x9-agent-instructions Skill
3★ repoPractical Agent Skills for source-backed research, controlled browser sessions, idea critique, and reliable AI workflows.
grusin-db/framework-skill-authoring/framework-skill-authoring Skill
no license3★ repoMeta-skill that turns any framework's source into a self-contained Agent Skills family — usable later from just the installed package (e.g. in Databricks Genie Code or any AI coding agent).
drsh4dow/grug-brain-skill/.agents/skills/skill-creator Skill
3★ repoA simplicity-first skill for reviewing architecture, refactors, APIs, and testing strategy without adding unnecessary complexity.
AlexFischman/mcp-skill-creator-agency/mnt/skills/skill-creator Skill
2★ repoSkillCreatorAgency: an Agency Swarm starter that connects a “skill creator” agent to the Claude Skills MCP backend for managing and testing local/remote skills.
kongyo2/kongyo-skill-creator/kongyo-skill-creator Skill
2★ repoLeonMelamud/claude-code-security-review/.github/skills/skill-creator Skill
no license2★ repoAI-powered security audit skill for code changes with false positive filtering. Based on Anthropic's claude-code-security-review.
Renol1/skill-creator-pro Skill
2★Production-grade skill creator for Claude Code, Cursor, Codex and more. 21 patterns, self-learning training loop, cross-platform.
harshitsinghbhandari/domain-expansion/.agents/skills/skill-creator Skill
no license2★ repoCollection of Claude agent skills (code audits, LLM councils, PR review, resume tooling) installable individually via npx skills add.
rodrigolagodev/skill-creator Skill
2★Expert guidance for creating high-quality Agent Skills following the open standard (https://agentskills.io)
Solonnikov/agent-skills/skills/agent-skill-creator Skill
no license2★ repoPublic collection of agent skills — reusable capabilities, prompts, and workflows for Claude Code and other agentic coding tools.
sleetcrash/sleetcrash-skills/skills/toolify Skill
1★ repoCurated Claude Code skills. First up: toolify, a session-mining skill that turns the work you just did into reusable tools.
Modellix/modellix-skill/.agents/skills/skill-creator Skill
1★ repoAn Agent Skill of Modellix documentation and capabilities reference.
zwbao/skill-creator-pro Skill
1★Pro-grade lifecycle for creating, testing, versioning, and iterating on Claude Code skills — fuses Anthropic's skill-creator, lovstudio's pro-skills, and superpowers' writing-skills.
knack-run/knack-starter-pack/skills/first-skill-template Skill
1★ repoFree MIT Claude Code skills: clarify rough ideas and scaffold your first skill.
arcjet/skills/.agents/skills/skill-creator Skill
1★ repoSkills to make your AI coding agent an Arcjet security expert.
valorisa/Claude-Skill-Creator/skill-creator Skill
1★ repoA meta-skill for Claude Code (Anthropic CLI) that guides you through creating, testing, and iterating custom skills via structured conversation. Designed with LLM Council methodology: interview intent, generate draft, smoke-test immediately, refine until satisfied. A skill is just a markdown file — this tool embraces that simplicity.
princekrz/skills-builder Skill
no license1★Universal toolkit for building, discovering, reviewing, testing, and securing Claude skills. 7 scripts, 10 reference docs, 6 workflows. Includes quality scoring, auto-testing, prompt injection detection, quarantine-based install, and conflict detection.
jajupmochi/agent-harness/.claude/skills/new-hook Skill
1★ repoLinlin's curated AI agent harness configuration: workflow rules, skills, hooks, plugin recommendations, tooling preferences, and project templates. Loadable into any new project so a fresh /init can pick the relevant subset.
morzecrew/agent-skills/.agents/skills/skill-creator Skill
1★ repoCollection of agent skills
Conradgui/skill-creator-pro/skills/skill-creator-pro Skill
1★ repoEngineer production-ready Codex skills with behavior contracts, quality lint, and forward testing / 将 Skill 构建从 vibe prompting 变成工程化流程
Skill creator anthropics 20260226
MatrixFounder/Universal-skills/archive/skill-creator-anthropics-20260226 Skill
1★ repoCollection of high-leverage "Meta-Skills" designed to upgrade AI Agents from simple chat bots to autonomous engineers
Jylhis/skills/meta/skill-creator-lang Skill
no license1★ repoCurated Agent Skills marketplace for Claude Code, Codex, and Google Antigravity.
jtmthf/skills/.agents/skills/skill-creator Skill
no license1★ repoPersonal Agent Skills
Betswish/meta-skill-generator Skill
1★Generate production-ready SKILL.md files for Codex, Claude Code, Cursor, and other agent-skill workflows.
lirantal/skill-creator-extra/skills/skill-creator-extra Skill
1★ repoA skill that creates skills, relies on the official skill-creator and then some!
Metaverse-Cloud/EngageLab-Skills/.agents/skills/skill-creator Skill
1★ repoEngageLab-Skills
krkn-s/ai-workspace/.pi/skills/skill-creator Skill
1★ repoKRKN's Personal AI workspace • Skills / Prompts / ...
eagerworks/skills/.agents/skills/skill-creator Skill
1★ repoA collection of portable agent skills by Eagerworks — install via the skills.sh CLI for Claude Code, Cursor, Copilot, Codex, Amp, and more.
udohjeremiah/skills/.agents/skills/skill-creator Skill
1★ repoReusable Agent Skills for AI coding agents, focused on software engineering workflows.
zjy1020/claude-code-skills/skills/agent-skill-creator Skill
0★ repoclaude 自己常用的技能
aditya-m-bharadwaj/APIskillBuilderSkill/.claude/skills/api-skill-builder Skill
0★ repoA Claude Code skill that turns a third-party REST API into a safe, single-file Python CLI + matching runtime skill — credential never enters AI context, every mutation is classifier-gated (read / mutating / destructive / billable / financial / privilege) and audit-logged.
pengqianhan/NZ-life-skills/skills/nz-skill-creator Skill
0★ repoPractical New Zealand life skills for international students, especially Chinese newcomers, with official-source-backed guides for Codex and Claude Code.
kongyo2/markdown-skill-creator/markdown-skill-creator Skill
no license0★ repolawwu/skills-marketplace/{{cookiecutter.repo_name}}/plugins/{{cookiecutter.plugin_name}}/skills/skill-creator Skill
no license0★ repoCookiecutter Template for an Agent Skills Marketplace
Steven-Herrera/skill-creator-supplement Skill
no license0★John-Dekka/advanced-skill-builder/advanced-skill-builder Skill
0★ repoAn interactive guide for creating production-ready Claude skills through structured dialogue. Whether you're building your first skill or enhancing an MCP integration, this skill walks you through requirements gathering, planning, YAML frontmatter generation, instruction writing, testing, and iterative refinement.
PalladiumGroupLtd/claude-chat-skills/chat-skill-creator Skill
no license0★ repoWorkflow for creating sandbox compatible skills in Claude
Tingtao-code/skill-intake Skill
0★Vet a candidate skill / MCP / prompt before adopting it into Claude Code/Codex — overlap, fit, weight & safety checks, then a one-shot install.