plain-english
Skill1 starsChanged 16 days agoMITView on GitHub
Extracts the signal from bloated or evasive text: the bottom line, every claim with its strength marked, what it does not say, and what it implies without claiming. Adds nothing from outside the source. Use for "what is this actually saying", "cut through this", "TL;DR this", or on pasted marketing, legal, policy, or AI-generated writing.
15 other projects use a version of this.
Adopted by
Most recently taken up by
- niksavis/handilytheir own version, first seen Nov 8, 2026
- niksavis/basiclytheir own version, first seen Nov 7, 2026
- sorawit-w/agent-skillstheir own version, first seen Nov 1, 2026
- meowdiocre/meowpitheir own version, first seen Oct 27, 2026
- fedemelo/agent-configtheir own version, first seen Oct 17, 2026
- MounirDhahri/skillstheir own version, first seen Oct 13, 2026
Used with
Projects with it also use find-skills.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
VercelSkillfind-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Also in 3 projects that use this
Other versions15 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 15
- b1rdmania/claude-plain-english-skill★ 46same name
- Tradelord223/plain-english★ 14same name
- fedemelo/agent-config★ 4same name
- naticorrea-03/claude-code-skills★ 4same name
- Harish-here/maddog★ 3same name
- meowdiocre/meowpi★ 3same name
- CorySchulz/skills★ 2same name
- JoseArron/skills★ 2same name
- niksavis/basicly★ 2same name
- niksavis/handily★ 2same name
5 more not listed
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 repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- global/CLAUDE.mdCLAUDE.md
- babysit-prsSkill
- claudeSkill
- copy-editSkill
- debriefSkill
- delegateSkill
- full-sendSkill
- geminiSkill
- launch-summarySkill
- merge-masterSkill
- mockupSkill
- phillipSkill
- phillip-syncSkill
- pointsgpt-supportSkill
- review-prSkill
- skill-editSkill
- sync-prsSkill
- ui-walkthroughSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 7 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 15 edited versions we read keep these.
7
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 26
Never: write anything before the bottom line.
- Distinctive: line 54
Every claim gets exactly one label.
- Distinctive: line 71
If a sentence carries two separately falsifiable claims, split it into two bullets.
- Distinctive: line 96
Never: add context, examples, definitions, or corrections from your own knowledge, however right.
- Distinctive: line 99
Keep an undefined domain term as the source wrote it.
- Distinctive: line 104
Do not: add a cause the source only implies by adjacency.
- Distinctive: line 113
When. "Soon", "in due course", "in the coming weeks" are absences, not dates.
- Very specific: line 145
Write the bottom line as Nothing actionable.
“Common” is measured: kept by at least 6 in 10 of the 15 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Plain English
Pull the signal out of the noise. Do not rewrite the text. Select from it.
The reader is someone who has to act on this text. The bottom line is the consequence for the reader, not the topic of the text.
The guarantee
Add nothing. Drop nothing that changes the reader's take.
Never carry a hedge, bound, condition, scope limit, or attribution forward as its own text. It rides along as part of the claim it modifies.
Output contract
Never write anything before the bottom line. No preamble, no "Here is the breakdown".
**Bottom line:** <the one thing that matters to this reader, or "Nothing actionable.">
**What it claims**
- **<strength>:** <the claim, with its hedge, number, and scope attached>
<optional: one sentence naming what the hedge costs, only when the label alone does not show it>
**Not stated**
- <what a reader needs and does not get>
**What it implies but never claims** <- only when this gap exists
- <the reading the text induces, and the device that induces it>Rules for the shape:
- Length tracks real content, not input length. Three sentences of substance produce three bullets. A 2,000-word post with one claim gets one. Never pad a section to make the output look proportionate to the input.
- Drop Not stated only when nothing material is missing. That is rare.
- Drop What it implies but never claims whenever there is no gap. That is common, and an empty section is worse than none.
Strength labels
Every claim gets exactly one label.
| Label | Means | Trigger words |
fact | Asserted flatly, checkable now | none needed |
hedged | Asserted with an escape hatch | the Hedges row under What must ride along |
attributed | Someone else's assertion, not the author's | the Attribution row under What must ride along |
promise | A future act the author commits to | will, plans to, by <date>, in the coming weeks |
opinion | A value judgment, not falsifiable | best, critical, exciting, industry-leading |
Judgment calls:
- Label by what the sentence does, not by its verb. "We will consider it" is
hedged, notpromise, because the committed act is only consideration. - A hedged promise is
hedged. The escape hatch wins. - Attribution outranks the rest. "The vendor says it cuts costs 40%" is
attributed, even though the inner claim is a fact. - If a sentence carries two separately falsifiable claims, split it into two bullets. Never merge two claims under one label.
What must ride along with a claim
Attach each of these to its bullet, in plain words.
| Carry | Words and test | |
| Hedges | may, might, could, should, expects, aims to, is designed to, we believe, we expect, estimated, projected, planned, targeted | <!-- lint-style: ignore --> |
| Attribution | according to, X says, reportedly, per, sources indicate, said | |
| Bounds | about, roughly, up to, at least, at most, more than, fewer than. "up to 40%" is not "40%". "at least 7 days" is not "7 days". | |
| Quantifiers | all, most, many, some, few, none, only | |
| Conditions and exceptions | if, unless, until, provided that, subject to, except, excluding, other than | |
| Negation | not, no, never, cannot | |
| Time and scope limits | as of <date>, in <year>, so far, in <region>, for <group>, on <platform> | |
| Numbers, units, dates, currency, versions, names, percentages | copied verbatim. Convert nothing, round nothing. |
Judge these by meaning. essentially and effectively are hedges when they mean
"close but not exactly". only is a bound in "only 3 users" and filler in "just
click here". significant is a claim when the source gives the number and vague
when it does not.
Add nothing
- Never add context, examples, definitions, or corrections from your own knowledge, however right. If the source is wrong, say the source claims it and put the correction nowhere.
- Keep an undefined domain term as the source wrote it. Do not gloss it. If it blocks meaning, that belongs under Not stated.
- Vagueness is data. "significant improvement" with no number is extracted as a claim with no number, so the reader learns the number is missing.
- Do not resolve an ambiguous pronoun by guessing.
- Do not add a cause the source only implies by adjacency. That belongs under What it implies but never claims.
Finding what is not stated
List only absences that block the reader's decision.
- Who acts, decides, or is accountable. Passive voice and "it was determined" are the tell.
- When. "Soon", "in due course", "in the coming weeks" are absences, not dates.
- How much. A percentage with no baseline, a gain with no denominator.
- Who is affected, and how they find out.
- What happens if it fails, or what the tradeoff cost.
- The comparison. "Faster" and "better" need a "than what".
List the absence, not a complaint about it. Write "Who decided", not "They conveniently avoid saying who decided".
What it implies but never claims
Include this section only when the text induces a belief it never asserts. Every bullet names the device, so the reader can check you.
- Adjacency posing as cause. "Sales fell. We changed the pricing." implies a link the text never claims.
- A hedge placed to read as a commitment. "We aim to ship in Q3" in a section headed "Q3 Deliverables".
- A statistic with no denominator. "Support tickets dropped 60%" implies fewer problems, and never says volume fell because a product was retired.
- An actor hidden by the passive so nobody appears responsible.
- Precision borrowed from an unrelated number. An exact figure next to a vague claim lends it false weight.
- A retracted commitment stated as progress. "We have decided to focus elsewhere" for work that was cancelled.
Name the device and stop. Never speculate about motive. Never write a bullet you cannot point at a specific sentence for.
When the source says nothing
What it claims comes out thin or empty, and Not stated comes out long.
Write the bottom line as Nothing actionable. followed on the same line by what
the text is doing instead, in one clause: announcing an announcement, restating
its own title, or promising detail later.
Do not manufacture claims to fill the section. An empty What it claims is a finding.
Reference
references/examples.md owns the worked cases.
Read it for these four calls only: may as granted permission versus may
as a hedge, the author as their own source, stacked qualifiers on one claim, and
an empty claims list.
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.

