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 isatimur/presentation-md --skill deck-design-judgegit clone --depth 1 https://github.com/isatimur/presentation-mdWrote 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/isatimur/presentation-md/deck-design-judge)<a href="https://agentmods.dev/skills/isatimur/presentation-md/deck-design-judge"><img src="https://agentmods.dev/badge/skills/isatimur/presentation-md/deck-design-judge/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/isatimur/presentation-md/deck-design-judge"><img src="https://agentmods.dev/badge/skills/isatimur/presentation-md/deck-design-judge.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 | $0.00042 | $0.01176 |
| Opus 5 | $0.00021 | $0.00588 |
| Sonnet 5 | $0.00008 | $0.00235 |
| Haiku 4.5 | $0.00004 | $0.00118 |
Grade A, and why
deck-design-judge 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 10d 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 — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deck design judge
Scorecard process: tier → locate → metrics → (shots) → rubric → merge. Leading words: scorecard, gate, evidence, self-critique.
Steps
1. Pick the tier
| Tier | Runs | When |
|---|---|---|
| T0 | deck_metrics.py only |
Fast sanity / CI |
| T1 (default) | Metrics + one LLM rubric judge | Almost always |
| T2 | T1 + screenshots the judge opens | Final visual QA |
| T3 | T2 + multi-model panel (median), via scripts/judge_panel.py |
Highest stakes — judge-panel.md |
Done when: tier is stated (default T1 if unsure).
2. Locate deck + brand tokens
Find the .html (or shots dir). Read nearby colors_and_type.css / token files /
logo SVGs. Note typeface, accent, radii, flat vs elevated.
Swiirl / Untitled-UI lineage is flat — elevation shadows are a gate.
Workspace: <deck-dir>/.design-eval/<timestamp>/.
Done when: deck path and flat/elevated rule are recorded.
3. Metrics (every tier)
python3 <skill>/scripts/deck_metrics.py <deck.html> \
[--tokens <colors_and_type.css>] -o <workspace>/metrics.json
Swiirl HTML (optional — swiirl-design is internal brand-token tooling; skip this step if you
don't have it):
python3 <swiirl-design>/scripts/check_brand_html.py <deck.html> \
-o <workspace>/brand_check.json
Interpret flags with metrics.md.
Done when: metrics.json exists and every gate id is listed.
4. Screenshots (T2 / T3 only)
bash <skill>/scripts/render_slides.sh <deck.html> <workspace>/shots/
Open each PNG — overflow, clipping, accent area, 3-second rule. No Chrome → fall back to T1 and note it.
Done when: every shot reviewed, or T1 fallback noted.
5. Rubric judge
Read rubric.md. Score all 10 dimensions 0–5 with
evidence (slide #, metric, quoted headline). Prefer the lower anchor when
merely “fine.” T3: run the real panel — scripts/judge_panel.py <deck.html> --metrics <workspace>/metrics.json --out <workspace>/judge.json (median per
model; errored models excluded, not zeroed) — see judge-panel.md.
What ships with it
22 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- .gitignore 62 B
- assets/scorecard-template.md 1.3 KB
- ATTRIBUTION.md 1.4 KB
- references/calibration.md 569 B
- references/judge-models.json 1.4 KB
- references/judge-panel.md 8.9 KB
- references/metrics.md 5.8 KB
- references/rubric.md 9.4 KB
- scripts/_isolate_slides.py 3.0 KB runs code
- scripts/brand_tokens.py 2.1 KB runs code
- scripts/deck_metrics.py 22 KB runs code
- scripts/judge_panel.py 22 KB runs code
- scripts/render_slides.sh 2.3 KB runs code
- scripts/scorecard.py 19 KB runs code
- scripts/tests/fixtures/brand.tokens.json 297 B
- scripts/tests/fixtures/sample.deck.json 889 B
- scripts/tests/test_brand_tokens.py 1.6 KB runs code
- scripts/tests/test_deck_json.py 3.3 KB runs code
- scripts/tests/test_deck_metrics.py 14 KB runs code
- scripts/tests/test_isolate_slides.py 2.1 KB runs code
- scripts/tests/test_judge_panel.py 20 KB runs code
- scripts/tests/test_scorecard.py 18 KB runs code
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.
- 10d ago First seen · 131 lines · 42 tokens per session scan A 5846a5444bf6
deck-design-judge is a skill published in the GitHub repository isatimur/presentation-md (4 stars, last pushed 1mo ago), licensed MIT. It adds 42 tokens to every session and 1,176 once invoked, about $0.0002 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-slides
Create stunning, animation-rich HTML presentations from scratch or by converting PowerPoint files. Use when the user wants to build a presentation, convert a PPT/PPTX to web, or create slides for a talk/pitch. Helps non-designers discover their aesthetic through visual exploration rather than abstract choices.
goose-graphics
Portable visual skill pack for the Agent Skills ecosystem (Claude Code, Claude Desktop, Claude Cowork, Claude Design, Goose, Cursor, Codex). Discovers community-published styles + formats via the gooseworks CLI, runs an extract-style workflow on reference images, and exports rendered PNGs via Playwright.
motion-graphics
Use when the user wants a short, design-led motion graphic where motion is the message: kinetic typography, stat or number count-up, chart/data-viz hit, logo sting, brand lockup, lower-third, callout, social overlay, animated headline/tweet/news item, motion poster, or quick captured-page highlight. Usually under 10s…
create-workflow-diagram
Create FigJam/Miro-style workflow diagrams as high-quality PNG images from plain-text workflow descriptions. Renders beautiful HTML diagrams with connected nodes, arrows, and labels, then screenshots them for sharing.
create-chatgpt-mockup
Render pixel-accurate ChatGPT mobile (iOS) screen mockups in light mode from a thread JSON. Supports user text bubbles, user image attachments, assistant markdown prose, citation chips, the OpenAI spiral logo, the Apps-SDK GPT chip in the composer, and three header styles (model-tag, plain title, "Get Plus"). Fixed…
create-imessage-mockup
Render pixel-accurate iMessage screenshot mockups (DM or group) from a thread JSON. Supports minimal, with-keyboard, and full iPhone 15 Pro frame variants. Outputs HTML + PNG.