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.
npx skills add GreatMark/fable-field-guide-skills --skill design-directionsgit clone --depth 1 https://github.com/GreatMark/fable-field-guide-skillsWrote 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.
[](https://agentmods.dev/skills/greatmark/fable-field-guide-skills/design-directions)<a href="https://agentmods.dev/skills/greatmark/fable-field-guide-skills/design-directions"><img src="https://agentmods.dev/badge/skills/greatmark/fable-field-guide-skills/design-directions/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/greatmark/fable-field-guide-skills/design-directions"><img src="https://agentmods.dev/badge/skills/greatmark/fable-field-guide-skills/design-directions.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00167 | $0.01318 |
| Opus 5 | $0.00084 | $0.00659 |
| Sonnet 5 | $0.00033 | $0.00264 |
| Haiku 4.5 | $0.00017 | $0.00132 |
Grade A, and why
design-directions 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 12d 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.
How it starts
The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Design Directions
The user has "unknown knowns": criteria they can only articulate when they see something. Give them things to react to, cheaply, before touching real code.
Workflow
- Confirm scope from context: what surface is being designed (page, panel, component, report) and what data it shows. Do not ask more than one clarifying question; guessing reasonably is part of the point. If it turns out the user wants a technical feasibility spike, say this skill does not apply and build the spike instead.
- Build 4 deliberately different directions by default (fewer only if the user asks; at one direction this skill's divergence machinery is moot — say so, and either just build it or propose a second contrasting take). Different means different layout philosophy, density, hierarchy, or interaction model — not one design with four color schemes. Before building, put all directions on one shared set of axes (layout philosophy, density, hierarchy, interaction model) and state each direction's position on them; every pair of directions must sit at opposite ends of at least one axis.
- Deliver as self-contained HTML: inline CSS/JS, realistic fake data
hardcoded inline (in the language of the conversation), no build step,
no network calls. Aim for high-fidelity visuals — the user should react
to the design, not the sketchiness; static is fine, and add only the
minimum JS that makes the core loop legible (e.g. a ticking timer for a
timer app). When interaction itself is the core loop (charts, drag,
filtering), prefer rendering one frozen mid-interaction moment (tooltip
already open, panel mid-drag) over scripting it; script only what a
static frame cannot convey. Include a viewport meta; default to desktop
web unless the surface is obviously mobile — then wrap in a phone-width
(390px) frame. For non-web surfaces (CLI/TUI, email, native), still
deliver HTML but simulate the medium honestly — monospace grid and
16-color palette for a terminal — and never show an effect the real
medium cannot render. Default to one file per direction (a single file
with a direction switcher only if the user asks). Write files to a
scratch directory under the project root (
./design-directions/; in a non-project directory, confirm where to put them first), never into the real app; do not commit them. In a git repo, exclude the directory:n=design-directions/; p="$(git rev-parse --git-path info/exclude 2>/dev/null)" && { mkdir -p "$(dirname "$p")" && { grep -qxF "$n" "$p" || echo "$n" >> "$p"; }; } 2>/dev/null || echo "note: could not exclude — leaving it untracked, never staging it" - Name each direction with a short memorable label and one sentence on the idea behind it, so the user can say "the second one, but denser".
- Self-check before showing: render each file (headless browser or screenshot tool if available) and verify it holds up at the target viewport — no layout breakage, all fake data visible, no console errors. Then the divergence gate: if the screenshots shown side by side would read as tweaks of one design, rebuild the weakest until they don't. If no renderer is available, downgrade explicitly: run static checks (tags balanced, inline JS parses, no external URLs, viewport meta present) and tell the user the files are unrendered — ask them to report any breakage on first open. Fix what you find before presenting; a broken first render defeats the purpose.
- Present for reaction: screenshot each direction and embed the images in
your reply (fall back to
open design-directions/*.htmlif you cannot render), and ask what to keep and what to kill. Then iterate on the chosen direction, or combine elements on request.
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.
- 12d ago First seen · 97 lines · 167 tokens per session scan A bf5f4babe394
design-directions is a skill published in the GitHub repository GreatMark/fable-field-guide-skills (9 stars, last pushed 1mo ago), licensed MIT. It adds 167 tokens to every session and 1,318 once invoked, about $0.0008 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 2026-08-31.
Other skills, from other repositories
frontend-ui-engineering
Builds production-quality, accessible, responsive user-facing UIs. Use when building or modifying interfaces and pages, creating components, implementing layouts, meeting WCAG accessibility requirements, managing state, or when the output needs to look and feel production-quality rather than AI-generated.
skillshare-ui-website-style
Skillshare frontend design system for the React dashboard (ui/) and Docusaurus website (website/). Use this skill whenever you: build or modify a dashboard page or component in ui/src/, style or layout website pages or custom CSS in website/, create new React components for the dashboard, add pages to the dashboard…
ai-ui-generation
AI-assisted UI generation patterns for json-render, v0.app, Google Stitch, Bolt Cloud, and Cursor workflows. Covers prompt engineering for component and full-stack app generation, review checklists for AI-generated code, design token injection, refactoring for design system conformance, and CI gates for quality…
awesome-accessibility-audit
An accessibility audit for web interfaces and markup, based on WCAG rules for people using keyboards, screen readers, or other assistive technology. It reviews structure, labels, focus, contrast, motion, and dynamic content.
obsidian-theme-dev
CSS/SCSS development patterns for Obsidian themes. Load when working with theme.css, SCSS variables, or CSS selectors.
opt-ui-install
Install or upgrade @reopt-ai/opt-ui in a consumer project, wire app.css, or add a Block/Surface page template. Auto-branches by current install state. Triggers on "opt-ui install", "opt-ui init", "opt-ui setup", "opt-ui upgrade", "opt-ui update", "opt-ui app.css", "opt-ui block", "opt-cli block add", "opt-ui surface"…