- Home
- Skills
- Writing & Content
- Human-Voice & Copywriting Pack
Works with the AI Agents you already use
Human-Voice & Copywriting Pack
This pack turns both problems into measurements an agent can check and fix, then adds a one-week retro so you learn from what you published.
$19
Human-Voice & Copywriting Pack
Example session with this skill installed
Scan bad.md for stock AI phrasing in English/Japanese, check launch copy for unsupported claims and hype, and calculate the 7-day conversion delta from retro_ok.json.
- Read your context and instructions
- Compiled the human-voice & copywriting
Real output from running the packaged scripts on small fixtures.
1. node ai_tells_scanner.mjs bad.md
$ node ai_tells_scanner.mjs bad.md
bad.md: score 18 (threshold 6) FAIL
+1 line 1 stock-phrase - delve / dive into: "delve into"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - opening cliche: "In today's fast-paced"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - it is important to note: "it's important to note"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - hype adjective: "game-changer"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - unlock the power: "unlock the power"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - filler adjective or verb: "robust"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - filler adjective or verb: "leverage"
fix: Say the concrete thing instead (a number, a name, a before/after).
+1 line 1 stock-phrase - filler adjective or verb: "holistic"
fix: Say the concrete thing instead (a number, a name, a before/after).
+2 line 1 rule-of-three - 4 three-item lists (e.g. "you plan, build, and ship faster").
fix: Keep the one or two items you can back with evidence; vary the shape of the rest.
+2 line 1 low-burstiness - Sentence lengths barely vary (variation 0.21, mean 70 chars over 10 sentences).
... (5 more lines)
(exit 1)
2. node launch_copy_gate.mjs bad.md
$ node launch_copy_gate.mjs bad.md
bad.md: 7 finding(s)
[warn] line 1 generic-substitution - A competitor could paste this unchanged: "The powerful all-in-one platform for everyone"
fix: Replace it with something only you can say: a number you measured, a file name, a before/after.
[error] line 3 ungrounded-number - Figure without evidence: "Boost your productivity 10x with our revolutionary tool."
fix: Say how it was measured (date, n, method) or delete the figure.
[error] line 3 ungrounded-number - Figure without evidence: "97% of users love it."
fix: Say how it was measured (date, n, method) or delete the figure.
[error] line 1 multiple-cta - 3 different calls to action: buy now | start free trial | contact us
fix: Pick one action for this page and demote the others to plain text.
[error] line 3 hype-guarantee - Absolute promise: "guarantee"
fix: Replace it with what you actually tested and what you do not cover.
[warn] line 3 hype - Exaggeration: "10x"
fix: Cut it or replace it with the measured fact behind it.
[warn] line 3 hype - Exaggeration: "revolutionary"
fix: Cut it or replace it with the measured fact behind it.
(exit 1)
3. node retro_delta_calc.mjs retro retro_ok.json (7-day retro)
$ node retro_delta_calc.mjs retro retro_ok.json
7-day retro (click rate): improved
before 0.12 [0.1013, 0.1416] n=1000
after 0.18 [0.1574, 0.205] n=1000
delta 0.06 (50% relative), z=3.757, p=0.0002
next: a different single variable than "headline"; need 1923 per arm (~26 days)
(exit 0)
4. node retro_delta_calc.mjs gate gates_bad.json (release blocked)
$ node retro_delta_calc.mjs gate gates_bad.json
pre-publish gate: BLOCK (1/7)
[pass] claims_grounded - checked by running the gate and reading the output
[missing] numbers_traceable - Gate not reported.
[unverified] single_cta - Marked pass but no evidence. [U]
[fail] no_exaggeration - two guarantee words remain
[missing] links_and_assets_checked - Gate not reported.
[missing] mobile_and_print_checked - Gate not reported.
[missing] owner_and_rollback_set - Gate not reported.
(exit 1)
Connects securely to your tools. The creator never sees your data.
About this skill
Stop taking the agent's word for it
Readers spot machine-sounding copy in the first paragraph, and launch posts lose trust when the numbers have no source. This pack turns both problems into measurements an agent can check and fix, then adds a one-week retro so you learn from what you published.
Each skill ships one self-contained Node.js script (Node 18+, standard library only). It prints findings or results with a fix, and exits non-zero when the gate fails. Add --json for machine-readable output.
The 3 skills
1. human-voice-technical-copy scores AI-sounding prose in English and Japanese:
- stock phrases, rule-of-three lists, flat sentence rhythm (burstiness)
- bold overuse, repeated abstract nouns, dash overuse
- includes an AI-tells catalog with replacements
2. evidence-grounded-launch-post gates launch posts and landing-page copy:
- the substitution test: lines a competitor could paste unchanged
- numbers with no evidence nearby
- several competing calls to action, guarantees and hype words
3. pre-publish-and-7day-retro checks seven pre-publish gates and computes the 7-day retro:
- a gate counts only with evidence; missing gates block release
- before/after rates with Wilson intervals, z test and a verdict
- sample-validity warnings and a one-variable plan for the next test with its sample size
Built for agents and for review
- Single file per skill, standard library only: copy one script anywhere and run it
- Works when the skill is loaded through the Agensi MCP: write the script to a temporary file, run it, delete it
- No network access, no shell commands, no environment reads
- Exit code and counts you can quote as evidence
What is in the ZIP
- README.md
- shared/verification-principles.md
- skills/evidence-grounded-launch-post/scripts/launch_copy_gate.mjs
- skills/evidence-grounded-launch-post/SKILL.md
- skills/human-voice-technical-copy/references/ai-tells-catalog.md
- skills/human-voice-technical-copy/scripts/ai_tells_scanner.mjs
- skills/human-voice-technical-copy/SKILL.md
- skills/pre-publish-and-7day-retro/scripts/retro_delta_calc.mjs
- skills/pre-publish-and-7day-retro/SKILL.md
The AI-tells score is a heuristic, not an authorship detector; read the text aloud after it passes.
The evidence check looks for markers near a number; it cannot tell whether the evidence is true.
A before/after comparison is weaker than a split test. The calculator says so when it can see the problem.
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
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
Requires