language-review
Skill4 starsChanged 6 months agoMITView on GitHub
Lints user-facing copy in the active language profile against site-config terminology rules, surface-specific tone (article vs cold email vs LinkedIn post), and language-specific style rules (punctuation, plurals, diacritics, smart quotes). Use after editing files containing user-facing strings — UI labels, marketing copy, articles, legal text, outbound messages. Returns structured findings; does NOT rewrite text and does NOT verify factual claims.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- LARi-UQAC/ResearchToolstheir own version, first seen Nov 15, 2026
- kaiohenricunha/dotbabeltheir own version, first seen Nov 1, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- LARi-UQAC/ResearchTools★ 14same name
- kaiohenricunha/dotbabel★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repositoryOne other file it gives its agents.
- articleSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 56
When invoking the agent, pass:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
language-review — surface-aware copy linter
This skill is a thin entry point that delegates to the copy-reviewer
agent in .claude/agents/copy-reviewer.md with the right context loaded.
When to invoke
The bundled post-edit-nudge.mjs hook fires this skill automatically
after edits to known user-facing files. The hook only nudges — the main
session decides whether to actually run a review based on whether the
edit touched copy or was pure logic.
You can also invoke manually:
/language-review # review uncommitted changes
/language-review src/pages/Foo.tsx # focus on one fileWhat the agent reviews
For every edit, the agent checks:
- Terminology — every
terminology.preferredandterminology.bannedrule insite-config.yaml. - Voice consistency — does the copy hit
voice.toneadjectives and avoidvoice.avoiditems? - Surface-specific rules — different rules apply to articles vs landing pages vs LinkedIn posts vs cold emails. The active language profile defines the surface table.
- Language quirks — diacritics, plurals, smart quotes, gendered agreement (whatever applies in the active profile).
- Pricing drift — if
pricingblock is set in config, prices in copy must match. - Phasing dates drift — if
phasing_datesblock is set, date claims must match. - Factual claims — flagged for human verification, not auto-verified.
What the agent does NOT do
- Doesn't edit files. Review only. The main session applies fixes.
- Doesn't verify factual claims. Flags them.
- Doesn't review non-user-facing strings (code identifiers, config values, vendor names, English comments in a Polish project, etc.).
- Doesn't rewrite for taste alone — respect the author's voice.
- Doesn't produce a polished "improved version" for fact-bearing content. Point fixes only.
Inputs
When invoking the agent, pass:
- The file paths to review (or a diff range).
- The active language profile path
(
.claude/skills/article/language-profiles/<profile>.md). Note:examples/language-profiles/is reference-only for contributors. The runtime reads the installed copies under.claude/skills/article/language-profiles/. - The relevant
site-config.yamlsections (terminology,voice,pricing,phasing_dates,disclaimer_template).
The agent returns findings in the structure documented in
.claude/agents/copy-reviewer.md.
Surface detection
The agent infers the surface from the file path:
| Path pattern | Surface |
*/blog/*.md, */articles/*.md, content/posts/* | Article |
*/pages/*.{tsx,jsx,vue,astro}, */components/landing/* | Landing page |
*/components/dashboard/* | App UI |
*/legal/*, *Regulamin*, *Privacy*, *Terms* | Legal page |
*/emails/*, *-email.{md,html} | |
*linkedin*, *social*, *post-*.md | Social post |
Each surface has its own ruleset in the active language profile (see
.claude/skills/article/language-profiles/english.md for the
structure).
Output
The agent always returns findings in a fixed structure:
- Quick verdict (one sentence)
- Language issues (must-fix / should-fix / nits)
- Claims to verify (flagged, not verified)
- Risky or unsupported claims (promotional overreach, legal exposure, misleading scope)
- Key point fixes (only when grammar is broken or fact is wrong)
The main session decides what to apply. The agent never edits.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

