pencil-skill AGENTS.md

pencil-skill AGENTS.md is an instructions file for Codex, OpenCode from Nisus74/pencil-skill. It costs 3,976 tokens per session, scanned A, original, MIT.

Project instructions for an unofficial plugin that helps coding agents work with Pencil design files, which use the .pen format, through the Pencil MCP server.

In plain words
What is it for?
Orienting agents in the repository and guiding work on Pencil designs across coding platforms.
Why use it?
It gives agents the project purpose, file structure, naming rules, and platform-specific setup they need to make consistent changes.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/nisus74/pencil-skill/agents-md
Clone the repo
git clone --depth 1 https://github.com/Nisus74/pencil-skill

Made for: Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for pencil-skill AGENTS.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/nisus74/pencil-skill/agents-md.svg)](https://agentmods.dev/instructions/nisus74/pencil-skill/agents-md)
Your own site
<a href="https://agentmods.dev/instructions/nisus74/pencil-skill/agents-md"><img src="https://agentmods.dev/badge/instructions/nisus74/pencil-skill/agents-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 3,976 This file is loaded in full into every session.
When invoked 3,976 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.03976 $0.03976
Opus 5 $0.01988 $0.01988
Sonnet 5 $0.00795 $0.00795
Haiku 4.5 $0.00398 $0.00398

Measured 4d ago against content hash 085ac995b2ae, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

pencil-skill AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 297 lines

How it starts

The opening of the file, as written. The whole thing — 297 lines — stays where its author put it; the contents beside it link to each section on GitHub.

pencil-dev-skill: Project Context

This is the canonical project-context file. All AI coding tools (Claude Code, OpenAI Codex, Cursor, etc.) should read this file for project context. Platform-specific files (CLAUDE.md) are thin pointers to this file.


Project Purpose

Unofficial community plugin. This project is not affiliated with or endorsed by the Pencil.dev team. For the Pencil editor, MCP server, and official documentation, visit pencil.dev.

This repository is a standalone, platform-agnostic AI coding skill plugin that teaches AI coding tools how to work with pencil.dev design files (.pen format) via the Pencil MCP server.

Core artifact: skills/pencil-design/SKILL.md, the platform-agnostic skill content. Platform adapters: .claude-plugin/plugin.json, .cursor-plugin/plugin.json, .codex-plugin/plugin.json, and gemini-extension.json are the minimum files required by each platform's installer; they exist only so users on those platforms can run a one-line install command. They are not the substance of the project.


Naming Conventions

Three names appear in this project, each scoped to a different layer:

Name Scope Where it appears
pencil-skill GitHub repo name Repo URL, clone URL
pencil-dev-skill Plugin package name plugin.json, marketplace listings
pencil-design Skill name SKILL.md frontmatter, skill activation triggers

This is intentional: the repo is the deliverable, the plugin is the package, and the skill is the capability the AI invokes.


Repository Structure

skills/pencil-design/              # The platform-agnostic core
  SKILL.md                         # The skill: YAML frontmatter + instructions (v0.8.0)
  references/                      # On-demand references loaded by the skill
    mcp-tools.md                   # Cookbook for all 9 MCP tools + composite recipes
    states.md                      # Component states + screen-level fault states + onboarding/settings states
    flows.md                       # Transitions between screens (modal, validation, back-stack, onboarding, settings, search)
    accessibility.md               # ARIA, focus order, APCA, ARIA live regions, keyboard shortcuts, WCAG 2.2
    modern-patterns.md             # Container queries, fluid type, AI-UI, animation timing, command palette, perceived perf
    pencil-cli.md                  # Full Pencil CLI reference + When CLI vs MCP table
    pen-schema.md                  # .pen file JSON schema reference (v2.14)
    batch-design-grammar.md        # batch_design JavaScript API (Insert/Copy/Update/Replace/Move/Delete/SetVariables/Generate/FindEmptySpace)
    advanced-canvas.md             # v2.14 canvas features: shader fills, mesh gradients, script nodes, ellipse arcs/donuts, prompt/context nodes
    component-anatomy.md           # Reading component structure: slots, descendants paths, state activation
    composition-patterns.md        # Compound components, slot design, variant naming, status workflow
    file-architecture.md           # Cover frame, section regions, hierarchical naming, multi-.pen layouts
    forms.md                       # Submit behaviour, validation, error display, autofill, mobile inputs
    interactions.md                # Keyboard, focus, hit targets, loading timing, destructive actions, URL-as-state
    visual-hierarchy.md            # Six levers, eye-flow patterns, whitespace, density strategy
    layout-patterns.md             # Hero variations, feature sections beyond three-card grid, pricing, dashboards, settings, list-detail, empty pages (cited 2025/2026 exemplars)
    iteration-patterns.md          # Failure-mode rescues (too busy/sparse/generic/un-premium), self-critique gate, reference-image translation, three-iteration limit
    microcopy.md                   # Voice/tone framework, action-specific CTAs, error message anatomy, empty/success/confirmation copy, loading copy, localisation
    mobile-patterns.md             # Safe areas, sheets vs modals, sheet detents, gestures, haptics, tab bars, native conventions per platform
    iconography.md                 # Stroke weight, sizing, semantic icons, accessibility (aria-hidden vs accessible name), family consistency
    performance-design.md          # Network budgets, Core Web Vitals (LCP/CLS/INP), virtualisation, image and font optimisation, theme-color
    industry-patterns.md           # 8 industry families with 15-20 rules each + completeness pressure tests for SaaS / Website / Mobile
    data-viz.md                    # 25-chart selection matrix, colour-blind palettes (Okabe-Ito, ColorBrewer, Viridis), dashboard tiles, anti-patterns
    style-catalogue.md             # 30+ named UI styles (menu) organised by family with mood, when-to-use, anti-pattern, exemplars
    colour-palettes.md             # 40+ palette recipes (menu) tagged by industry/mood; recipes point to Tailwind/Radix/IBM Carbon/Material 3/Apple HIG
    font-pairings.md               # 30+ typography pairings (menu); recipes point to Google Fonts/Vercel/GitHub/commercial foundries
    codex-tools.md                 # OpenAI Codex tool name mappings
  design-system/                   # Optional design-system reference templates (one level deep)
    README.md                      # Agent loading guide
    CUSTOMISING.md                 # Plain-English guide for non-technical editors
    accessibility.md               # Project a11y standards (WCAG/APCA, keyboard, screen reader)
    empty-states.md                # Per-surface empty state catalogue with copy templates
    file-architecture.md           # Project .pen file structure and naming conventions
    forms.md                       # Form conventions (validation, error display, save patterns)
    micro-interactions.md          # Per-interaction motion specs
    navigation.md                  # Primary nav patterns, workspace switcher, mobile tab bar
    onboarding.md                  # First-run experience (sample-data vs blank slate)
    search.md                      # Search shape (instant / submit / hybrid), filters, URL state
    visual-style.md                # Project's chosen style identity (style + palette + font picks)
  examples/                        # 15 worked examples with real MCP tool sequences (one level deep)
    example-login-screen.md        # Greenfield auth screen
    example-import-library.md      # Import .lib.pen library + instantiate components
    example-error-screen.md        # 404 + offline page pair
    example-form-flow.md           # Multi-step signup with email verification
    example-component-deep-dive.md # Full read→understand→instantiate cycle
    example-style-selection.md     # Catalogue (style + palette + fonts) → SetVariables → tokens commit → starter components
    example-settings-page.md       # Settings with sidebar nav, autosave + explicit-save for high-stakes
    example-dashboard.md           # KPI cards + chart tile + recent-activity table
    example-marketing-page.md      # Marketing page avoiding three-card grid (asymmetric hero, bento features)
    example-mobile-app.md          # Mobile app with bottom tab bar, sheet detents, safe areas, haptics
    example-data-visualization.md  # Multi-chart dashboard with colour-blind-safe palettes
    example-onboarding-flow.md     # Three-step onboarding with progress, skip, sample-data routing
    example-component-variants.md  # Complete Button family with all variants and states
    example-pricing-table.md       # Three-tier pricing with highlighted recommended tier
    example-file-cover-and-sections.md # .pen file with Cover frame, section regions, hierarchical naming

# Platform install adapters (required by each platform's installer)
.claude-plugin/plugin.json         # Claude Code plugin manifest
.claude-plugin/marketplace.json    # Claude Code marketplace listing (single-plugin marketplace)
.cursor-plugin/plugin.json         # Cursor plugin manifest (Cursor 2.5+)
.codex-plugin/plugin.json          # OpenAI Codex plugin manifest
gemini-extension.json              # Gemini CLI extension manifest

# Project context files
AGENTS.md                          # This file — canonical, platform-agnostic
CLAUDE.md                          # Thin pointer to AGENTS.md (for Claude Code)
HARNESSES.md                       # Cross-platform skill capability matrix (frontmatter, directories, substitution)

# Public-facing
README.md
LICENSE

# Repo hygiene
.gitignore                         # Includes secret patterns
.gitattributes                     # Cross-platform line-ending normalization
.gitleaks.toml                     # Secret-scanning config

# Quality tooling
tools/
  skill-lint.py                    # OWASP Agentic Skills Top 10 lint (CI + pre-commit)
  test_skill_lint.py               # 40 unit tests for skill-lint
  requirements.txt                 # pip deps for Dependabot

# Documentation
SECURITY.md                          # Security policy + OWASP AST compliance map (repo root)
docs/
  CONTRIBUTING.md
  CODE_OF_CONDUCT.md
  CHANGELOG.md

# GitHub repo configuration
.github/
  PULL_REQUEST_TEMPLATE.md
  ISSUE_TEMPLATE/
  CODEOWNERS
  dependabot.yml
  workflows/
    secret-scan.yml                # gitleaks on push + PR
    skill-lint.yml                 # skill-lint + unit tests on push + PR
.pre-commit-config.yaml            # Local gate: skill-lint + gitleaks + hygiene

Read the full file on GitHub · 297 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 297 lines · 3,976 tokens per session scan A 085ac995b2ae

Subscribe to this mod's changes

pencil-skill AGENTS.md is an instructions file published in the GitHub repository Nisus74/pencil-skill (22 stars, last pushed 1mo ago), licensed MIT. It adds 3,976 tokens to every session, about $0.0199 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.