full-review
Skill0 starsChanged 3 months agoMITView on GitHub
Runs a structured multi-pass review of human-facing professional copy and presents findings as a before/after diff table with reasoning, lists what was deliberately left out, and waits for explicit sign-off before changing anything. The passes are AI tells, a register-supplied positioning pass, honesty and distortion (catching claims silently inflated OR softened), tone and positioning, voice register, and fact verification. Calibration comes from a register file (job-seeker by default; publication included) and an optional one-page profile, so the same review serves résumés, editorials, or a house style; add a register for anything else. Use at any stage of refining or expanding copy with AI — not just at the end — when reviewing or workshopping a resume, cover letter, portfolio page, case study, proposal, or bio, when the user asks to "review my copy", "do a full review", "run the full pass", or polish writing for hiring managers or clients.
10 other projects use a version of this.
Adopted by
Most recently taken up by
- arcasilesgroup/ai-engineeringtheir own version, first seen Nov 8, 2026
- Smaui94/ClaudeSkillCollectiontheir own version, first seen Nov 8, 2026
- blamechris/chroxytheir own version, first seen Nov 8, 2026
- fmind/dottheir own version, first seen Nov 8, 2026
- fmind/dotfilestheir own version, first seen Nov 8, 2026
- porada/domfilestheir own version, first seen Nov 1, 2026
Other versions10 projects use a version of this.
Word for word · 1
- cabbagecachekid/neon-jetpack★ 0100% same
Edited · 0
None found.
Wrote their own version · 9
- arcasilesgroup/ai-engineering★ 65same name
- fmind/dot★ 13same name
- fmind/dotfiles★ 13same name
- blamechris/chroxy★ 8same name
- porada/domfiles★ 7same name
- u7-u7/dev-skills★ 4same name
- garthcodes/claude-skills★ 2same name
- denismrvoljak/claude-scholar★ 1same name
- Smaui94/ClaudeSkillCollection★ 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 repository2 other files it gives its agents.
- ai-written-checkSkill
- cringe-checkSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 6 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 8 edited versions we read keep these.
6
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Use it at any stage of refining copy with AI, not only the final pass — any time you've drafted, expanded, or edited something with Claude …
- Very specific: line 20
If the author declines or the question can't be asked, run job-seeker with no profile — the pack's original behavior.
- Very specific: line 21
If the register file itself cannot be found (this skill was copied standalone, without registers/), run the job-seeker rules that are …
- Distinctive: line 27
Run every pass explicitly and report under each heading, even if a pass is clean ("Pass 2: no issues").
- Distinctive: line 40
Never silently upgrade scope: If an edit makes a claim bigger than the source — broader ownership, larger team, more authority, an earlier …
- Distinctive: line 43
Match the record, not the memory: When raw material already sounds inflated, treat that as a question to verify, not a fact to polish.
- Distinctive: line 49
Options get a recommendation: When you offer choices, present them as options *and* state your recommendation with honest reasoning (name …
- Distinctive: line 50
Use the author's own words: If they gave you a phrase, use it verbatim.
“Common” is measured: kept by at least 6 in 10 of the 8 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
Full Review
A disciplined, multi-pass review for high-stakes human-facing copy — resumes, cover letters, portfolio pages, case studies, proposals, bios. This is the orchestrator of the red-pen pack: it runs the full process and delegates the two focused checks (ai-written-check, cringe-check) rather than restating them, then adds the passes only it owns.
Use it at any stage of refining copy with AI, not only the final pass — any time you've drafted, expanded, or edited something with Claude and want to make sure the result still says what you meant and claims only what's true.
The first draft of any change is the starting position for a conversation, not the final proposal. Expect 2–4 iterations per piece.
Register and profile (load before any pass)
The mechanism below is register-independent; the calibration is not. Resolve it in this order, once per review:
- If
.red-pen/profile.mdexists in the working directory, load it. Its frontmatter names aregister:; loadregisters/<that>.mdfrom the pack root (two directories above this skill's folder in the plugin or cloned repo). The profile's voice test, never-phrases, scope guard, and identity rules bind every pass. - Otherwise ask once: "Which register —
job-seeker,publication, or a path to a profile?" Do not ask again in the same review. - If the author declines or the question can't be asked, run
job-seekerwith no profile — the pack's original behavior. - If the register file itself cannot be found (this skill was copied standalone, without
registers/), run the job-seeker rules that are inline below and say so in the first line of the report; do not invent a register.
State the resolved register in the first line of the report. If the profile has a scope guard and the draft falls outside it, say so and stop; do not drag the draft into a register it doesn't belong to.
The six passes (run in order)
Run every pass explicitly and report under each heading, even if a pass is clean ("Pass 2: no issues").
- AI tells — delegate to the
ai-written-checkskill. Apply it directly; do not restate its catalog here. (If it isn't installed, see the 10-second fallback inreferences/PASSES.md.) - Register pass — supplied by the register file's
## Pass 2 rule. Forjob-seekerthis is operator-vs-academic (ships vs. observes; every "research about X" pairs with a "so that <decision shipped>"). Forpublicationit is the mechanical house-style pass. Report it under the name the register gives it. - Honesty / distortion — does any claim now say more, or less, than the author's record? Never silently change the scope of a claim in either direction. The register's
polarity.distortionsays which direction AI editing damages this artifact (inflate for job-seeker, deflate for publication) and its## Distortion watchlistlists the tells — see the anti-distortion rules below. - Tone / positioning — delegate to the
cringe-checkskill. Apply it directly; do not restate its dimensions here. - Voice register — does each paragraph hold one register, or swerve from concrete-and-human to abstract industry-speak? The swerve is the tell.
- Fact verification — for every verifiable fact (dates, conference and company names, titles, metrics, technical terms), push for a source rather than trusting recall. Memory is unreliable; sources beat memory. Flag any unsourced fact as "verify before sending." Claims the register lists under
## Needs-no-source claims(opinion and analytical framing, for publication) are reported as opinion — no source required, not as gaps.
Anti-distortion rules (Pass 3, the heart of this pack)
These skills exist because AI editing can quietly change the size of a claim the author never approved — amplify it in application copy, soften it in argument. Never inflate, never deflate. Polarity is emphasis, not exclusion. Changing the size of a factual claim against the author's record is a violation in every register, in either direction; the register only says which direction to look hardest and supplies the style-level tells (softening, swagger) that are register-specific. The rules below are the job-seeker (inflation) form and remain in force under every register:
- Never silently upgrade scope. If an edit makes a claim bigger than the source — broader ownership, larger team, more authority, an earlier origin — STOP, flag it, and ASK. Do not inflate unprompted.
- Scope-amplifier watchlist: "solo," "founded," "created," "established," "built from scratch," "ran/led the program," "owned," "didn't exist before," "the entire." Each one is a claim that must match what actually happened.
- Read "solo" conservatively. "Solo" usually means individual contribution on a specific project, not sole founder of a function or program. Never expand it past per-project scope without explicit confirmation.
- Match the record, not the memory. When raw material already sounds inflated, treat that as a question to verify, not a fact to polish.
Output format (non-negotiable)
- Before/after diff table with three columns: Before · After · Why. Every proposed change is a row. No silent edits.
- What was kept out. List what you deliberately did NOT include or change, and why — especially anything disclosure-sensitive you withheld. As important as what you put in.
- Options get a recommendation. When you offer choices, present them as options and state your recommendation with honest reasoning (name the trade-off, not just the upside). A bare list makes the author decide cold.
- Use the author's own words. If they gave you a phrase, use it verbatim. Don't paraphrase their voice into yours. Keep industry vocabulary intact.
The sign-off rule
Never change the file until the author explicitly says "land it" (or equivalent). Review proposes; the author disposes. Draft and iterate in a working/scratch location; only move locked or near-locked text into the live/published location.
If the author pushes back with feedback you think weakens the writing, say so and explain why — honest disagreement over reflexive agreement. If they're right and you missed something, own it directly ("I missed this; my earlier pass failed because…"). Don't soften, defend, or bury the miss.
Composition
Passes 1 and 4 are the ai-written-check and cringe-check skills — install all three of red-pen together for the complete review; both read the same resolved register. references/PASSES.md carries the full detail for the four passes this skill owns (2, 3, 5, 6) plus a brief fallback for 1 and 4 if those skills aren't present, so a review can still run standalone. Registers live in registers/; the profile template and examples in profile.template.md and examples/.
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.

