You... actually think those are good. WOW.
Ok, so, virtually every one of them is straight instructions with almost no notational support, no polysemanticity, no narrative weight, and distinctly lacking in designer-grade language of fine distinction. Here. I'll feed one to my ulta-basic prompt improver tool.
Original
## Django Unit Test Generator for Viewsets
```
I want you to act as a Django Unit Test Generator. I will provide you with a Django Viewset class, and your job is to generate unit tests for it. Ensure the following:
1. Create test cases for all CRUD (Create, Read, Update, Delete) operations.
2. Include edge cases and scenarios such as invalid inputs or permissions issues.
3. Use Django's TestCase class and the APIClient for making requests.
4. Make use of setup methods to initialize any required data.
Please organize the generated test cases with descriptive method names and comments for clarity. Ensure tests follow Django's standard practices and naming conventions.
```
Improved:
```
Generate a Django REST Framework test suite for the provided ViewSet as if you’re shipping it into a real codebase tomorrow: start by inferring the concrete API surface (router basename, list/detail routes, allowed HTTP methods, serializer(s), queryset/model, permissions/auth, filters/search/ordering, pagination) directly from the ViewSet code, then write a `tests/test_<resource>_viewset.py` module using `django.test.TestCase` `rest_framework.test.APIClient` with a clean `setUp()` (or `setUpTestData()` when safe) that creates: (1) a “happy-path” user with required perms, (2) an unauthorized/anonymous client, (3) a user lacking perms, and (4) minimal valid model instances via factories or explicit model creation when factories aren’t shown. Implement full CRUD coverage with separate, descriptive test methods for create/list/retrieve/update(partial full)/delete, and make each test assert **status code response shape DB effects** (object count deltas, field updates, soft-delete behavior if present). Add edge-case tests driven by common DRF failure modes: invalid payloads (missing required fields, wrong types, boundary lengths), permission denials (401 vs 403 correctness), object-not-found (404), method-not-allowed (405 for disabled actions), validation errors (400 with field-level messages), uniqueness constraints, and any custom `perform_create/perform_update`, `get_queryset`, or `get_serializer_class` branches you detect. If the ViewSet varies behavior by action/user, include scenario tests that prove those branches. Keep tests deterministic: freeze time if timestamps matter, avoid ordering flakiness by explicit ordering, and avoid relying on auto-generated IDs except where necessary. Organize with helper methods for URL building (e.g., `list_url()`, `detail_url(obj)`), payload builders (`valid_payload()`, `invalid_payload_*()`), and small inline comments explaining *why* the edge case exists. When details are ambiguous (router basename, required fields, permission class semantics), ask *one* focused clarification at the top; otherwise choose sane defaults and note the assumption in a short comment block. Output only the finished test module, ready to paste, with imports, clear naming, and zero placeholder “TODO”s.
**ViewSet Code**:
```
You see? Or something a bit more stark. Here's a rando-prompt:
Original:
```
I need a marketing campaign strategy focused on the below provided psychometric customer profile.
```
# Psychometric-Aligned Marketing Campaign Architect
```
Design a marketing campaign strategy that locks onto the psychological architecture of the target customer and builds outward from their internal drivers, not demographic proxies. Begin by reverse-engineering the provided psychometric profile—identify dominant motivations (e.g., status-seeking, security-oriented, novelty-driven, tribe-aligned), cognitive style (analytical vs. intuitive), risk tolerance, identity anchors, aspirational self-image, and friction triggers.
Translate these traits into five aligned campaign pillars:
1. **Narrative Core** – Craft the central story that mirrors how this audience sees themselves (or wants to). Use their value language. Encode identity reinforcement over product explanation.
2. **Emotional Leverage Points** – Map primary and secondary emotional drivers (pride, fear of loss, belonging, mastery, rebellion, stability, etc.) and specify how each appears in messaging, visuals, and calls-to-action.
3. **Message Framing Strategy** – Define tone register (authoritative, insurgent, intimate, data-driven, visionary), proof style (statistics, testimonials, symbolic authority, social proof density), and objection-preemption tactics based on their skepticism profile.
4. **Channel & Touchpoint Selection** – Recommend platforms and formats congruent with their attention patterns and cognitive load tolerance (short-form punch vs. long-form depth; community hubs vs. broadcast; email cadence vs. event spikes).
5. **Conversion Architecture** – Design the funnel logic in behavioral terms: what belief must shift at each stage? What micro-commitment reduces resistance? Where does identity affirmation occur?
Embed structural intelligence:
- Identify psychological “red zones” (language, imagery, or tactics that would repel this profile).
- Include segmentation variations if the psychometric data suggests sub-clusters.
- Provide sample headline angles and offer hooks calibrated to this audience’s motivational stack.
- Outline 2–3 campaign concepts (distinct strategic directions) and briefly compare their expected resonance and risk.
Deliver as a concise strategic brief structured for implementation—clear sections, tactical clarity, and explicit reasoning linking psychometric insight → marketing decision.
If any psychological inputs are missing or vague, ask targeted clarification questions before finalizing the strategy.
**Required Params**:
**Psychometric Profile (full detail)**:
**Product/Service**:
**Primary Business Goal (awareness, leads, sales, retention, etc.)**:
**Budget Range (optional but useful)**:
**Geographic Scope (if relevant)**:
```
Those prompts are... ideas for prompts. Sketches. Poorly done ones, at that.
The exception is the art prompts. Those look pretty decent. But all those stars? The bland leading the bland to the Midsdom of Crowds. They've only ever seen crap and think "slightly less crappy" = "good".