Prompt engineering · daily drops

Prompts that behave like hired staff, not toys.

Meevra builds, tests and curates prompt systems for every major model — Claude Projects, Custom GPTs, Gemini, GLM, Kimi, Qwen and DeepSeek. Copy, paste, get professional output. Every prompt is original, tested and versioned.

Tuned for
New drops added daily · 19 free prompts inside
18models covered
100%original & tested
0 €to start
The arsenal

Packs built for how you actually work

Not 10,000 copy-pasted scraps — focused systems with setup instructions, per-model variants and copy-ready prompts.

Free

Meevra Prompt Vault

The living library. A curated set of tested prompts across every category — business, SEO, code, image, analysis. New drops land daily. Sign in with Google to unlock copying.

all modelsdaily dropslibrary
Free with sign-in Enter vault
Best sellerLaunch −40%

Claude Projects — Business Ops

Twelve ready-to-import Claude Projects: operations analyst, offer strategist, weekly review facilitator, hiring screener and more. Each with system prompt, knowledge structure and first-message script.

ClaudeProjects12 systemsrubric-scored
25 € 15 € Get the pack
PopularLaunch −40%

Custom GPT Instructor Kit

Everything to build Custom GPTs people actually use: instruction templates that stop drift, knowledge-file structures, conversation starters that work, and a testing checklist before you publish.

OpenAICustom GPT9 templatesrubric-scored
20 € 12 € Get the pack
Smart buyLaunch −40%

Free Models, Pro Results

GLM, Kimi, Qwen and DeepSeek are free — most prompts don't use their strengths. This pack retunes proven workflows for each model's character: long documents, reasoning chains, code, analysis.

GLMKimiQwenDeepSeekrubric-scored
15 € 9 € Get the pack
For buildersLaunch −40%

Rules, Slash Codes & Agents

CLAUDE.md files, cursor rules, system prompts for autonomous agents, and the slash-command patterns power users lean on. Make your editor and your agents behave the same way every day.

CLAUDE.mdcursor rulesagentsrubric-scored
15 € 9 € Get the pack
NewLaunch −40%

Video & Motion Pack

Shot systems for Sora, Veo, Kling, Runway and Higgsfield: scene → camera → motion → light, with consistency anchors that keep a series identical.

SoraVeoKlingRunwayHiggsfieldrubric-scored
20 € 12 € Get the pack
For foundersLaunch −40%

Builder Prompts Pack

Paste-ready specs for Lovable, Bolt and Cursor, plus a .cursorrules generator that stops the agent fighting your architecture.

LovableBoltCursorrubric-scored
15 € 9 € Get the pack
Real-timeLaunch −40%

Grok Social Kit

Ride real-time X chatter: the Trend Angler turns noise into three shippable angles, the Hook Forge ranks 10 first lines by stop-power.

GrokX / Twitter2 systemsrubric-scored
15 € 9 € Get the pack
EverythingLaunch −49%

All-Access Bundle

Every pack, every model vault, every future drop for 12 months, and the full vault without limits. When a new model matters, the tuning lands here first. One payment, one place.

all packs16 vaults12 mo updates
82 € 49 € Get everything
How it works

From payment to first output in two minutes

01

Pick your pack

Choose the workflow you want to upgrade — or start free in the vault with a Google sign-in.

02

Pay with PayPal

Secure checkout, instant delivery. The download link and setup guide arrive the second payment clears.

03

Paste & run

Every prompt ships with its model, placement and first message. Copy, paste, get the result.

Daily drops

The library that never sleeps

Every day Meevra scans the model landscape — new releases, new behaviors, new workflows — and ships prompts that answer them. Vault members see them first.

Researched

Model updates and community patterns reviewed daily; only what survives testing gets published.

Versioned

Prompts carry versions. When a model changes behavior, the tuning is updated — bundle owners get it free.

Attribution-clean

Every prompt is written and tested in-house. No scraped lists, no stolen packs, no legal fog.

Model vaults

One vault per provider

Tap any model in the theater above and you land straight on its vault — everything we ship for that model, growing every day. 7 € each, or take the All-Access bundle and stop choosing.

Open a demo vault — see inside (test) ↓
The proof

Benchmarked, not believed

Every flagship system is scored against the same task run without it — same model, same settings, five runs each, judged on a fixed rubric. Two representative cases:

Data-to-Insight Analyst

Task: 30 rows of raw shop revenue. Baseline prompt used: "Analyze this data."

Baseline — no promptavg 1.0 / 5
The data shows an upward trend over the month, with some fluctuations along the way. Consider focusing on marketing strategies to boost sales even further.
Structure2
Evidence shown1
Hallucination guard0
Decision-ready1
With the Meevra systemavg 4.5 / 5
STRUCTURE — 30 rows · 4 columns · Sep 1–30 TOP FINDING — Wednesdays carry 2.1× median revenue (12 of the top-20 days are Wednesdays) DECISION — move the newsletter send to Tue 18:00 · metric to watch: Wed conversion (numbers cited from the data; nothing invented)
Structure5
Evidence shown4
Hallucination guard5
Decision-ready4

Builder-Ready Spec Prompt

Task: "make me an app for my barbershop" — pasted into an AI app builder.

Baseline — no promptavg 0.8 / 5
Great idea! A barbershop app could include features like online booking, a services gallery, team introductions and customer reviews. Would you like me to start with the homepage?
Completeness1
Checkability0
Scope control1
With the Meevra systemavg 4.7 / 5
PRODUCT — one-location booking app for a barbershop DATA — Barber(name, specialty) · Slot(time, barber_id, taken) · Client(name, phone) ACCEPTANCE — "double-booking the same slot is rejected" · "cancelled slot frees for re-booking" · "no data persists after cache clear unless saved" OUT OF SCOPE — payments, multi-location, staff payroll
Completeness5
Checkability5
Scope control4

Honest limits: in-house rubric runs on frontier models (5 runs per side, temperature 0.7), excerpts trimmed for display — the full protocol and raw runs ship inside the packs. Not peer-reviewed; we publish the method so you can rerun it yourself.

Questions

Fair answers

Can't I just ask AI to write prompts for me?

You can — and we do exactly that, then test, fix, restructure and re-test. The value isn't the text; it's the testing, the per-model tuning and the maintenance when models change. Free vault first, decide after.

Which models are covered?

Claude (Projects & chat), OpenAI GPT (chat & Custom GPTs), Gemini, GLM, Kimi, Qwen and DeepSeek. Image prompts cover Midjourney-class and Flux-class models.

How do I get my files after paying?

Checkout is PayPal; delivery is automatic — the download page opens immediately after payment and a link is sent to your email.

Do prompts stop working when models update?

Sometimes — that's why every pack includes updates. Bundle owners receive every revision for 12 months at no extra cost.

Why sign in with Google for the free vault?

One tap, no password to manage, and it lets us send you the daily drop. No spam, unsubscribe anytime.