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 tmj-90/gaffer --skill i18n-l10ngit clone --depth 1 https://github.com/tmj-90/gafferWrote 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/tmj-90/gaffer/i18n-l10n)<a href="https://agentmods.dev/skills/tmj-90/gaffer/i18n-l10n"><img src="https://agentmods.dev/badge/skills/tmj-90/gaffer/i18n-l10n/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/tmj-90/gaffer/i18n-l10n"><img src="https://agentmods.dev/badge/skills/tmj-90/gaffer/i18n-l10n.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 | $0.00108 | $0.01726 |
| Opus 5.5 | $0.00043 | $0.00690 |
| Sonnet 5.5 | $0.00022 | $0.00345 |
| Haiku 4.5 | $0.00011 | $0.00173 |
Grade A, and why
i18n-l10n 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 16d 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Internationalise and localise correctly
A hard-coded string or a hand-built plural is a bug that surfaces the day a second locale
ships. Externalise every user-visible string as a whole message with named arguments,
let CLDR data (through ICU MessageFormat and the Intl APIs or the language's
equivalents) decide plurals and formats, and build layouts that survive longer text and
right-to-left scripts.
Steps
- Use the repo's mechanism. Library (i18next, FormatJS/react-intl, vue-i18n,
Angular i18n, Lingui, gettext, ICU4J/ICU4X, .NET resources, Apple String Catalogs,
Android resources), message file layout, key scheme, locale detection and persistence.
Call
search_lore. Never add a second mechanism or a parallel strings file. - Externalise every string users can see or hear: labels, placeholders,
altandaria-labeltext, validation and error messages, emails, notifications, page titles. Keys describe meaning and location (ticket.actions.approve), never the English text. APIs return error codes and parameters; the client localises them. - Write translatable messages.
- Whole sentences with named arguments:
{name} approved {count, plural, one {# ticket} other {# tickets}}— never"Approved " + n + " ticket(s)". - Plurals use CLDR categories (
zero,one,two,few,many,other). English uses onlyone/other, but Polish needsfew/manyand Arabic all six, so never branch onn === 1in code.otheris always present.=0 {No tickets}is an exact match, distinct from thezerocategory. - Ordinals use
selectordinal("1st, 2nd"); gendered text usesselect. - Rich text uses the library's tag syntax (
<b>{name}</b>), not HTML concatenation. - Give translators a description for ambiguous short strings ("Open" verb or adjective?).
- Unicode MessageFormat 2 is stable in CLDR 47, but most libraries still use ICU MessageFormat 1 syntax; write what the repo's library parses.
- Whole sentences with named arguments:
- Format with the locale APIs.
Intl.DateTimeFormat(preferdateStyle/timeStyleover pattern strings),Intl.NumberFormat(includingstyle: 'currency'with an ISO 4217 code,'percent','unit',notation: 'compact'),Intl.RelativeTimeFormat,Intl.ListFormat,Intl.PluralRules,Intl.Collatorfor sorting,Intl.Segmenterfor truncating text by grapheme or word,Intl.DisplayNamesfor language/region names. Reuse formatter instances; they are costly to construct. - Money and time. Money is integer minor units plus the currency code; minor units
vary (JPY 0, USD 2, KWD 3) — take them from the formatter or ISO 4217, not a constant
100. Past instants are stored as UTC and displayed in the user's IANA time zone
(
Europe/Berlin, not "CET" or an offset). A future local event (a 09:00 meeting) is stored as local date-time plus IANA zone, so a DST rule change does not move it. UseTemporalwhere the runtime supports it, else the repo's date library; never parse locale-formatted strings back into dates. - Locale selection. BCP 47 tags (
pt-BR,zh-Hant-TW); negotiate from the user's saved preference, thenAccept-Language/navigator.languages, then the default; fall back through the chain (de-AT→de→ default). Never infer language from IP or country. Set<html lang>anddir, andlangon inline passages in another language (WCAG 3.1.1, 3.1.2). - Layout for expansion and direction. Allow roughly 30–40% longer text (more for
short labels) without clipping; no fixed-width buttons around text. Use CSS logical
properties (
margin-inline-start,inset-inline-end,text-align: start) and flex/grid, which mirror automatically underdir="rtl". Mirror directional icons (back/forward arrows), not universal ones (play, checkmark, logos). Isolate user-generated text inside messages (<bdi>or the library's bidi isolation) so an Arabic name does not reorder the sentence. - Keep the catalogue honest. Add new keys to the source-locale file in the same change; run the repo's extraction and missing/unused-key check; remove keys you orphaned. Translations arrive through the repo's workflow — do not machine-translate into production catalogues unless the ticket says to.
- Test (the
frontend-testingskill), per AC:- render under a pseudo-locale (accented, ~40% longer, bracketed:
[Ţîçķéţš ~~]) or a long real locale (de, fi) → no raw keys, no untranslated literals, no clipping; - plurals at 0, 1, 2, 5, 21 and 22 in a locale with
few/many(pl or ru); - dates, numbers and currency snapshot for two locales and a non-UTC time zone
(set
TZin the test run); dir="rtl"renders mirrored where the repo supports RTL. Record per-AC evidence with therecord-evidenceskill, then stop.
- render under a pseudo-locale (accented, ~40% longer, bracketed:
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.
- 16d ago Changed · +33 lines · +8 tokens per session af4e94756082
- 19d ago First seen · 65 lines · 100 tokens per session scan A 0ecc8416f5a8
i18n-l10n is a skill published in the GitHub repository tmj-90/gaffer (2 stars, last pushed 17d ago), licensed Apache-2.0. It adds 108 tokens to every session and 1,726 once invoked, about $0.0004 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 2026-09-28.
Other skills, from other repositories
bulwark-brainstorm
Role-based brainstorming with dual modes: --scoped (sequential Task tool, 5 roles) and --exploratory (Agent Teams peer debate, 4 roles). Use for feasibility assessment and idea validation.
plan-creation
Create structured implementation plans via a 4-role scrum team (Product Owner, Architect, Eng/Delivery Lead, QA/Critic) with optional Agent Teams peer debate mode.
anthropic-validator
Validates Claude Code assets (skills, hooks, agents, commands, MCP servers, plugins) against official Anthropic standards. Fetches latest docs dynamically and produces structured validation reports.
create-subagent
Generates single-purpose Claude Code sub-agents for use via the Task tool. Use when creating dedicated sub-agents, scaffolding agent definitions, or generating agents with diagnostics and permissions setup.
test-audit
Audit test suites for T1-T4 violations using AST analysis, mock detection, and multi-stage synthesis. Invoke when user asks to audit tests, check test quality, find mock violations, review test effectiveness, or inspect test suites for over-mocking. Triggers automatic rewrites when quality gates fail.
code-review
Comprehensive code review with distinct aspect based sections. Use when reviewing code, checking for security issues, finding type safety problems, auditing code quality, or when user asks to review code, PRs or changes. Three-phase workflow runs static tools, LLM judgment, and writes diagnostic log.