- Home
- Skills
- Writing & Content
- Remove AI Tells: Ranked Fixes, Proof Check, Voice (EN/IT)
More screenshots
Works with the AI Agents you already use
Remove AI Tells: Ranked Fixes, Proof Check, Voice (EN/IT)
Readers spot machine-written text in a few words, and once they do they stop trusting what it says.
Free
Remove AI Tells: Ranked Fixes, Proof Check, Voice (EN/IT)
Example session with this skill installed
Rewrite this follow-up to a recruiter so it doesn't sound like AI. Facts: I applied on 12 May, I ran 14 CRM rollouts in my last role, the largest for 900 users.
"Hi Maria, I hope this email finds you well! I wanted to follow up on my application for the Solutions Engineer role. I am excited to bring my robust experience in CRM implementations to your vibrant team. It's not just about technology, it's about people. Additionally, I believe my background serves as a testament to my ability to deliver pivotal results, ensuring customer success. Let me know if you'd like me to share more details. I hope this helps! Best, Alex"
- Read your context and instructions
- Compiled the remove ai tells
Rewritten email
Hi Maria,
I applied for the Solutions Engineer role on 12 May and wanted to check whether the team has started interviews. In my last role I ran 14 CRM rollouts, the largest for 900 users, so I can walk you through a similar case if useful.
Best,
Alex
Checker, before and after
python3 scripts/check_text.py draft.txt
Before: 0 errors, 13 warnings
| Line | Defect | Found |
|---|---|---|
| 2 | #12 chatbot residue | I hope this email finds you well |
| 2 | #8 brochure vocabulary | I am excited to, robust, vibrant, Additionally, testament, pivotal |
| 2 | #1 false contrast | It's not just about technology, it's |
| 2 | #9 inflated verb | serves as |
| 2 | #11 -ing tail | , ensuring |
| 2 | #15 same openings | three sentences in a row start with "I" |
| 3 | #12 chatbot residue | Let me know if you'd like me to, I hope this helps |
After: 0 errors, 0 warnings
What changed
- The request goes first: has the team started interviews.
- "Robust experience" and "pivotal results" became the facts you gave: 14 rollouts, the largest for 900 users.
- The false contrast, the greeting filler and the closing offers are gone; no fact was added beyond the three you provided.
Next steps
- Paste the email and send it.
- Say "build my voice profile" with 5 to 20 of your sent emails so future drafts follow your rhythm.
Connects securely to your tools. The creator never sees your data.
About this skill
Readers spot machine-written text in a few words, and once they do they stop trusting what it says. This skill keeps every fact and removes the patterns that give the text away. It works on anything you send or publish under your own name: emails, replies, cover letters, application answers, messages to recruiters or colleagues, posts, bios.
What it does
- Removes 16 AI tells, ranked by damage. False contrasts ("it's not X, it's Y"), punchline endings, run-ups ("here's the thing"), brochure words (delve, leverage, seamless), chatbot residue ("I hope this helps") and eleven more. The five worst are fixed even when they appear once; the weak ones only when they pile up.
- Proof check on every claim about you. An adjective like "proven" or "extensive" stays only if the same sentence holds the fact behind it. The swap test asks whether another candidate could write the sentence word for word; if yes, it gets tied to a fact or dropped.
- Never adds facts. Names, numbers and dates come only from you or your sources. If a sentence needs a missing fact, the agent asks.
- Messages that respect the reader. One purpose, one request, nothing the reader already knows. In replies the decision goes first.
- Voice profile. Say "build my voice profile" and paste 5 to 20 texts you really wrote. The agent writes a profile of about twenty lines (rhythm, openings, closings, what you never do), shows it to you and saves it as
voice-profile.mdafter your approval. Every later draft follows it. - Checker script.
scripts/check_text.pyflags dashes and the typical phrasings with line numbers, so the final check does not depend on the agent's attention. Python 3.8+, standard library only, no network.
English and Italian
Every defect lists typical phrasings in both languages, and the checker runs with --lang en, --lang it or --lang both.
What is inside
SKILL.md: rules and procedure.references/defects.md: the 16 defects with phrasings and fixes.references/example-voice-profile.md: a fictional profile showing the expected shape.scripts/check_text.py: the checker.README.md: install and usage.
Limits
- The checker finds phrasings, not shapes: triplets, repeated endings and contrasts split across two sentences are left to the agent's reading.
- WARN lines are candidates: quotes and literal uses of a flagged word stay.
- Two languages: English and Italian.
The checker finds phrasings, not shapes. WARN lines are candidates, not verdicts. English and Italian only.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 16 days ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Runs the bundled checker on a temporary draft file. Writes only voice-profile.md, after the user approves the profile. No network access.