say-it-once

say-it-once is a skill for Claude Code, Codex from ktwu01/benchmark-radar. It costs 57 tokens per session (869 once invoked), scanned A, original, MIT.

A procedure for finding repeated or unhelpful elements in a webpage, dashboard, README, or design brief. It checks the rendered page at desktop and phone widths rather than relying only on source code.

In plain words
What is it for?
It helps review charts, rankings, cards, panels, and other visible interface elements before opening a pull request.
Why use it?
Repeated facts take up space and can make readers wonder whether two copies differ. The procedure also catches labels or accessible names that would be lost when removing duplicates.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit It helps review charts, rankings, cards, panels, and other visible interface elements before opening a pull request.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/ktwu01/benchmark-radar/say-it-once
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.

Any agent
npx skills add ktwu01/benchmark-radar --skill say-it-once
Clone the repo
git clone --depth 1 https://github.com/ktwu01/benchmark-radar

Made for: Claude Code, Codex.

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 say-it-once

README.md
[![agentmods](https://agentmods.dev/badge/skills/ktwu01/benchmark-radar/say-it-once/github.svg)](https://agentmods.dev/skills/ktwu01/benchmark-radar/say-it-once)
Your own site
<a href="https://agentmods.dev/skills/ktwu01/benchmark-radar/say-it-once"><img src="https://agentmods.dev/badge/skills/ktwu01/benchmark-radar/say-it-once/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.

agentmods 80×15 button for say-it-once

Your own site · 80×15
<a href="https://agentmods.dev/skills/ktwu01/benchmark-radar/say-it-once"><img src="https://agentmods.dev/badge/skills/ktwu01/benchmark-radar/say-it-once.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 57 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 869 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00057 $0.00869
Opus 5.5 $0.00023 $0.00348
Sonnet 5.5 $0.00011 $0.00174
Haiku 4.5 $0.00006 $0.00087

Measured 13d ago against content hash 9af3ff014957, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

say-it-once 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 13d 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.

skills/say-it-once/SKILL.md · 57 lines

How it starts

The opening of the file, as written. The whole thing — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Say it once

Each fact a reader sees appears in one place on the screen. A second copy costs space above the answer and makes the reader check whether the two copies differ. design.md holds the rules; this skill is the procedure for applying them.

Steps

  1. Open the rendered surface in Chromium at 1450px and at 390px. Use the built site from a clean worktree, or the live site for an unchanged surface. Do not judge from source alone: most past duplicates were assembled at runtime.
  2. List every element above the main answer (chart, ranking, list), top to bottom. For each, write the one fact it carries.
  3. Group the elements by fact. A group with two or more members is a candidate. Also flag any element whose fact is unreadable alone, such as a bare number with no label (94.6 above a chart that already says Best reported score: 94.6).
  4. For each candidate, keep the copy in the place the reader looks for that fact, and remove the others. Before removing, check that the copy is not the only one that:
    • carries state or a control: keyboard, phone width, Back and Forward, the URL;
    • carries an accessible name. Move that text into aria-label, as eb1b872 did, instead of dropping it;
    • is unique evidence: provenance, a caveat no other element states, a citation.
  5. Machine copies are not duplicates. Meta description, og: tags, JSON-LD, feeds, and datePublished serve crawlers and previews (cbda65c). Leave them.
  6. Check order as well as count. A block shared across surfaces follows each surface's job (design.md, "A shared block follows the surface's job").
  7. Take before and after screenshots at the same viewport. For each removal, name in the commit message the element that already says it: "X already says Y".

Patterns from this repo's history

Commit Repeated element Kept
8741e16 (#298) Eyebrow, title, source chip and a second heading: four elements for two facts, 150px above the chart One title and one subline
cbda65c Eyebrow "Daily brief", a date line, fixed tags, and a clipped lede above its full text The title; the clipped text stays in meta and the feed
eb1b872 Number, "/ 100", bar, caption, and rubric mark for one score Number, bar, mark; the caption moves to aria-label
69a261d Owner in the title and in the byline Owner once, as "By"
7fa1725 A Code link that opens the same repository as the rail Skipped only when it opens the same kind of page on the same site
2c21c37 A dataset card that repeated the footer's Data link The footer link
887ae5f An info mark beside an already-blue link The blue
ca7c1cc (#976) "About {name}" under an <h1> naming the benchmark The <h1>

Read the full file on GitHub · 57 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. 13d ago First seen · 57 lines · 57 tokens per session scan A 9af3ff014957

Subscribe to this mod's changes

say-it-once is a skill published in the GitHub repository ktwu01/benchmark-radar (363 stars, last pushed yesterday), licensed MIT. It adds 57 tokens to every session and 869 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-10-27.

Related

say-it-once compared with skills from other repositories

say-it-once next to 6 skills from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
say-it-oncektwu01/benchmark-radarSkill363yesterdayMITA57Claude Code, Codex
ai-prompt-regression-testingsickn33/agentic-awesome-skillsSkill49ktodayMITA37Claude Code
evaluation-nemo-evaluatordavila7/claude-code-templatesSkill33ktodayMITA76Claude Code, Codex
evaluation-bigcode-evaluation-harnessdavila7/claude-code-templatesSkill33ktodayMITA70Claude Code, Codex
connect-agentlangwatch/langwatchSkill5.0ktodayApache-2.0A102Claude Code, Codex
prompt-optimizationlangwatch/langwatchSkill5.0ktodayApache-2.0A105Claude Code, Codex
dashboard-widgetslangwatch/langwatchSkill5.0ktodayApache-2.0A79Claude Code, Codex