New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Agent Instructions

by murphytrueman in murphytrueman/design-system-ops

Write the AGENTS.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request.

MITAuto-check passedAgent Workflows

Install Agent Instructions

skills CLI
$ npx skills add murphytrueman/design-system-ops --skill agent-instructions -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install murphytrueman/design-system-ops agent-instructions --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/murphytrueman/design-system-ops.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/agent-instructions .claude/skills/agent-instructions && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
agent-instructions
GitHub stars
203
Token cost
~2.3k tokens
SKILL.md length
986 words
Files
1
Skills in repo
36
Repo updated
First seen
Licence
MIT

At a glance

Write the AGENTS.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request.

  • Works in 5 steps: Inventory what an agent could be pointed… → The provenance rule → Write AGENTS.md → …
  • Tasks that involve Agent instruction files
  • SKILL.md covers Before you begin: verify…, Context, Boundaries and Configuration, plus 6 more sections
  • Calls npx

What it does

Agent Instructions is an agent skill from murphytrueman/design-system-ops. Write the AGENTS.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request. Triggers: AGENTS.md for our system, make it AI-navigable, agent instructions. Inventory: codebase-index.

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Agent instruction files and Design systems. The repository describes itself as: Claude Code skills for the work that keeps a design system alive. The licence is MIT.

When your agent uses it

  • Tasks that involve Agent instruction files
  • Tasks that involve Design systems

Example prompts

  • “/agent-instructions”

Requirements

  • Node.js
  • Pre-approved tools (allowed-tools): Read, Write, Grep, Glob, Bash(cat:*), Bash(find:*), Bash(head:*), Bash(ls:*)

Workflow steps

5 steps, taken from the step headings in SKILL.md.

  1. Inventory what an agent could be pointed at
  2. The provenance rule
  3. Write AGENTS.md
  4. Entry-point pointers (only when asked)
  5. Summarise in chat

What it can do on your machine

Read from SKILL.md and the folder at commit f167898. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Write
    • Grep
    • Glob
    • Bash(cat:*)
    • Bash(find:*)
    • Bash(head:*)
    • Bash(ls:*)

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • npx

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use npx, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Agent Instructions loads about 2.3k tokens when it runs. Until then it costs about 79 tokens; SKILL.md has 986 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~79
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from murphytrueman/design-system-ops at commit f167898, republished under its MIT licence (© murphytrueman). 986 words, ~2,256 tokens.

Download SKILL.mdSave it as .claude/skills/agent-instructions/SKILL.md (or your agent's skills folder).
name
agent-instructions
description
Write the AGENTS.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request. Triggers: AGENTS.md for our system, make it AI-navigable, agent instructions. Inventory: codebase-index.
allowed-tools
Read, Write, Grep, Glob, Bash(cat:*), Bash(find:*), Bash(head:*), Bash(ls:*)
references
../../knowledge-notes/ai-readiness.md, ../../knowledge-notes/output-discipline.md

Agent instructions

A skill for writing the file coding agents actually read: an AGENTS.md at the repository root that tells an agent where the system's parts live, which rules apply and where each rule comes from, how to check its work, and what not to do. It links to the machine-readable files other skills produce (.ai/index/, .ai/metadata/, decision pages, token files) rather than copying them, so there is one place to keep current.

Before you begin: verify references

Confirm that every path in this skill's frontmatter references: exists relative to this SKILL.md. If any is missing, stop: the install is incomplete, usually because a flattening installer (for example npx skills install) dropped the repo-root knowledge-notes/ directory. Tell the user to reinstall by a method in 1-INSTALL.md and run verify-install.sh from the install root. Proceed without the references only if the user explicitly says to, and then say in the output that it was produced without the pack's reference material.

Context

Agents working in a codebase read a small set of conventional files first: AGENTS.md, CLAUDE.md, .cursor/rules/, .github/copilot-instructions.md. Design system knowledge that isn't reachable from one of those is invisible at the moment it matters: the agent picks a component, types a colour, or writes a test from what it can see, and what it can see is props and values, not intent. The design system loses the argument it never got to make.

The fix is not more files in .ai/. It is one short file the agent will read, with pointers to the rest, and rules that are the team's own. Long instruction files get skimmed; unsourced rules get treated as suggestions or, worse, as policy the team never set.

Boundaries

This skill writes instructions; it doesn't build the inventory (codebase-index), the per-component metadata (metadata-schema-generator), the token docs (token-documentation) or the choosing-between pages (component-decision-tree). It links to those when they exist and names them as the next step when they don't. If the system has no components, tokens or docs to point at yet, there is nothing to instruct; say so.

Configuration

If .ds-ops-config.yml exists, follow the configuration-and-recurring knowledge note (../../knowledge-notes/configuration-and-recurring.md). This skill reads:

  • system.name, system.framework, system.styling, system.component_paths, system.tokens
  • agent_instructions.path — where to write the file (default AGENTS.md at the repo root)
  • agent_instructions.pointers — extra entry points to write, each a one-line file that says "read AGENTS.md" (default: none; ask)

Step 1: Inventory what an agent could be pointed at

Look before writing. Record the path of each thing found, or that it wasn't found:

  • Existing instructions: AGENTS.md, CLAUDE.md, .cursor/rules/*, .github/copilot-instructions.md, .windsurfrules, a docs/ai/ folder. If any exists, this skill updates it: keep what the files confirm, replace what they contradict, and say what changed. Never write a second file that competes with one the team already maintains.
  • Machine-readable outputs from this pack: .ai/index/ (inventory and graph), .ai/metadata/ (per-component props, a11y contracts), .ai/decision-trees/ or the markdown choosing pages, .ai/tokens/ if token-documentation wrote one.
  • The system itself: component source directories and the barrel export; the token source and its consumption form (custom properties, JS module, Tailwind theme, Sass); the docs site or Storybook URL from the README; CONTRIBUTING.md; CHANGELOG.md.
  • Enforcement that already exists: ESLint and Stylelint configs (a declaration-strict-value rule, no-restricted-imports, jsx-a11y), lint and test scripts in package.json, CI workflow names. These are the rules that are already true; the file should tell the agent to run them.
  • Written rules: contribution guidelines, an accessibility policy, a content or voice guide, decision records. Each is a source a rule can cite.
Show full SKILL.md (408 more words)Show less

Step 2: The provenance rule

Every rule in the file carries its source in brackets: a file path, a URL, a decision record id, or team when the user stated it in this session. A rule with no source is not written as a rule. If it's worth keeping, it goes in a final "Proposed, not yet policy" list with a one-line reason, for the team to accept or delete. Agents treat the file as the system's actual policy; an invented rule becomes policy the moment it's written down. See "Every figure and fact needs a source" in the output-discipline note.

Step 3: Write AGENTS.md

Keep it under about 150 lines. Real paths, real names, no placeholders except [confirm: …] where the team hasn't answered, listed at the top. Structure:

markdown
# Working with [System name]

Still to confirm: [list of `[confirm: …]`, or "none"]

## What this is
[Two sentences: what the system provides, who owns it, the docs URL.]

## Where things live
- Components: `src/components/` — exported from `src/index.ts`
- Tokens: `tokens/*.tokens.json` (DTCG), consumed as CSS custom properties from `src/styles/tokens.css`
- Component inventory and dependency graph: `.ai/index/` (run `codebase-index` to refresh)
- Per-component props and accessibility contracts: `.ai/metadata/`
- Choosing between components: `docs/choosing/` [or "not written yet"]
- Docs / Storybook: [URL]

## Use these, not those
- Use exported components from the package; don't re-implement or copy source. [CONTRIBUTING.md#local-components]
- Use semantic tokens (`--color-action-primary`, `--space-inset-md`), never raw values and never primitives (`--color-blue-500`). [tokens/README.md]
- When a component or variant doesn't exist: [the contribution route]. Don't build a local one. [CONTRIBUTING.md]

## Rules
[One line each, each with a source. Only what the team has actually written down or stated.]

## Accessibility
[The system's guarantee and standard, with source; where per-component contracts live; what stays the consumer's job.]

## Check your work
- `npm run lint` (ESLint: `no-restricted-imports`, `jsx-a11y`; Stylelint: `declaration-strict-value`)
- `npm test`
- [any design-system-specific check, from package.json]

## Don't
[Three to six concrete don'ts with sources, e.g. no `!important` on token-driven properties, no `jest.mock` of the design system package.]

## Proposed, not yet policy
[Unsourced rules the team should accept or delete, each with a one-line reason. Delete this section when empty.]

Write the content from Step 1's findings. Where a section has nothing to say (no .ai/metadata/, no accessibility policy), write the honest one-liner ("not written yet; run metadata-schema-generator") rather than inventing a contract.

Step 4: Entry-point pointers (only when asked)

Some tools read their own file first. If agent_instructions.pointers lists them, or the user asks, write each as a pointer, never a copy, so AGENTS.md stays the single source:

  • CLAUDE.md: "Read AGENTS.md before working with the design system."
  • .cursor/rules/design-system.mdc: frontmatter with alwaysApply: true and the same one line.
  • .github/copilot-instructions.md: the same one line.

If one of these already exists with its own content, append the pointer; don't replace the file.

Step 5: Summarise in chat

  • Headline: the file written or updated, and how many rules it holds with sources versus proposed
  • Files: every path written, and every existing file appended to
  • Linked, not copied: which .ai/ and docs artefacts the file points at, and which don't exist yet
  • Scope: the block from the output-discipline note: what was inspected, what wasn't, and every [confirm]

Quality checks

  • The file is under about 150 lines and every path in it exists in the repository
  • Every rule has a source in brackets; unsourced rules are in "Proposed, not yet policy" or absent
  • Token and component names in the examples are the system's real names, taken from the source
  • "Check your work" lists commands that exist in package.json, not aspirational ones
  • Machine-readable files are linked, never duplicated into the instructions
  • An existing instructions file was updated in place, and the summary says what changed
  • Nothing in the file describes a generic design system; a reader could tell which system it is from any section

© murphytrueman, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/agent-instructions of murphytrueman/design-system-ops.

Open the folder on GitHubat commit f167898

Compare with similar skills

Agent Instructions next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Agent Instructions compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Agent Instructions this skillmurphytrueman/design-system-ops203—~2.3kAutomated safety check: PassMIT
Agentic UI Developmentantdigital-ai/agentic-ui224—~555Automated safety check: PassMIT
Drydock Repo Harness SetupYeachan-Heo/oh-my-claudecode40k—~5.2kAutomated safety check: PassMIT
ReflectNikiforovAll/claude-code-rules141—~1.8kAutomated safety check: PassApache-2.0
Research Setupbrycewang-stanford/Auto-Empirical-Research-Skills4.5k—~1.2kAutomated safety check: PassCustom licence
Improve UI Audit and Plansibelick/ui-skills9.5k—~2kAutomated safety check: PassMIT

Similar skills

  • Agentic UI Development

    antdigital-ai/agentic-ui

    Guides development workflow, design system, scripts, and component inventory for @ant-design/agentic-ui.

    224 GitHub stars~555 tokensUpdated yesterday
    Frontend & DesignAuto-check passed

  • Drydock Repo Harness Setup

    Yeachan-Heo/oh-my-claudecode

    Sets up a shared repo environment for people and agents: CLAUDE.md, a CONTEXT.md glossary, ADRs, standards and a design-system folder, and audits drift with --check.

    40k GitHub stars~5.2k tokensUpdated today
    Agent WorkflowsAuto-check passed

  • Reflect

    NikiforovAll/claude-code-rules

    Analyze a Claude Code session for "wrong-turn" moments (corrections, retries, waste, reversals, dead-ends) and produce an interactive HTML dashboard with copy-able recommendations (CLAUDE.md rules…

    141 GitHub stars~1.8k tokensUpdated 6 days ago
    Agent WorkflowsAuto-check passed

  • Research Setup

    brycewang-stanford/Auto-Empirical-Research-Skills

    Interactive setup wizard that configures a new project for the social-science-research plugin.

    4.5k GitHub stars~1.2k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed

  • Improve UI Audit and Plans

    ibelick/ui-skills

    Audits one product surface against its own design evidence and writes self-contained implementation plans for another agent, without touching product source.

    9.5k GitHub stars~2k tokensUpdated yesterday
    Frontend & DesignAuto-check passed

  • Fast Agent Design

    evalstate/fast-agent

    A skill your agent uses to generate well-branded interfaces and assets for fast-agent, either for production or throwaway prototypes/mocks/etc.

    3.9k GitHub stars~220 tokensUpdated today
    Frontend & DesignAuto-check passed

More from murphytrueman/design-system-ops

All 36 skills in this repo

  • AI Component Description

    murphytrueman/design-system-ops

    Write a six-section prose description (purpose, props, anti-patterns, composition, accessibility, examples) for a Figma component's description field so LLMs read it via MCP.

    203 GitHub stars~4.7k tokensUpdated 14 days ago
    Auto-check passed

  • Change Communication

    murphytrueman/design-system-ops

    Write release notes, a migration guide and a team announcement for a design system change that is already decided, scaled to its impact.

    203 GitHub stars~3.4k tokensUpdated 14 days ago
    Auto-check passed

  • Codebase Index

    murphytrueman/design-system-ops

    Generate machine-readable index files in .ai/index/ (component inventory, uses/usedBy graph, stats) for AI agents.

    203 GitHub stars~4.7k tokensUpdated 14 days ago
    Auto-check passed

  • Codemod Generator

    murphytrueman/design-system-ops

    Generate tested jscodeshift/postcss codemods for design system migrations: token renames, prop renames or removals, import paths, component swaps.

    203 GitHub stars~4.9k tokensUpdated 14 days ago
    Auto-check passed

  • Component API Validator

    murphytrueman/design-system-ops

    Audit prop APIs across a component library: naming consistency, boolean/default patterns, type coverage, exported types, breaking changes between versions.

    203 GitHub stars~4.3k tokensUpdated 14 days ago
    Auto-check passed

  • Component Decision Tree

    murphytrueman/design-system-ops

    Write "choosing between" pages (docs/choosing/) that route an intent to the right component via narrowing questions; YAML trees on request.

    203 GitHub stars~4.9k tokensUpdated 14 days ago
    Auto-check passed

Categories

Questions about Agent Instructions

What does Agent Instructions do?

Write the AGENTS.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request. Agent Instructions is an agent skill from murphytrueman/design-system-ops.md that tells coding agents how to use this design system: where things live, sourced rules, how to check work, what not to do; Claude, Cursor or Copilot pointers on request.

When should I use Agent Instructions?

Agent Instructions fits situations like: tasks that involve Agent instruction files; tasks that involve Design systems.

How do I install Agent Instructions in Claude Code?

Run `npx skills add murphytrueman/design-system-ops --skill agent-instructions -a claude-code`. Or copy the skill folder (skills/agent-instructions in murphytrueman/design-system-ops) into .claude/skills/agent-instructions in your project. Claude Code loads it when a task matches its description.

How do I install Agent Instructions in Codex?

Run `npx skills add murphytrueman/design-system-ops --skill agent-instructions -a codex`. Or copy the skill folder (skills/agent-instructions in murphytrueman/design-system-ops) into .agents/skills/agent-instructions in your project. Codex loads it when a task matches its description.

Can I use Agent Instructions in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add murphytrueman/design-system-ops --skill agent-instructions -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/agent-instructions, .gemini/skills/agent-instructions, .github/skills/agent-instructions and .opencode/skills/agent-instructions in your project.

What does Agent Instructions need to run?

Going by SKILL.md and its folder, Agent Instructions needs the command-line tools its instructions call (npx). Our summary lists: Node.js. Its frontmatter pre-approves these tools: Read, Write, Grep, Glob, Bash(cat:*), Bash(find:*), Bash(head:*), Bash(ls:*).

Does Agent Instructions access the network?

SKILL.md contains no URLs. Its commands use npx, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Agent Instructions safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Agent Instructions use?

Agent Instructions is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Agent Instructions use?

About 2.3k tokens (SKILL.md is roughly 9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Agent Instructions?

Skills that share tags, products or a category with Agent Instructions: Agentic UI Development (antdigital-ai/agentic-ui, 224 stars), Drydock Repo Harness Setup (Yeachan-Heo/oh-my-claudecode, 40k stars), Reflect (NikiforovAll/claude-code-rules, 141 stars) and Research Setup (brycewang-stanford/Auto-Empirical-Research-Skills, 4.5k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Agent Instructions?

murphytrueman (a GitHub user) maintains it in murphytrueman/design-system-ops, which has 203 GitHub stars. The repository holds 36 skills in this directory. The repository was last updated on September 24, 2026.

Source: murphytrueman/design-system-ops on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.