{{ }}PromptVault

24 prompts

Clusters raw interview quotes into themes with evidence.

Synthesize these {{n}} customer interviews about {{topic}}.

Output:
1. Themes (name + supporting quotes, cited by interviewee)
2. Contradictions between interviewees
3. Jobs-to-be…
researchClaude{{3}}#synthesis#user-research
10 uses · Aug 26, 2026Open →

Template for building few-shot classification prompts.

You are a {{domain}} classifier. Classify each input into exactly one of: {{labels}}.

Examples:
{{examples}}

Now classify the following. Reply with ONLY the label, nothing else.
…
systemGPT-4oClaudeLlama{{4}}#classification#few-shot
14 uses · Aug 26, 2026Open →

Simplifies jargon-heavy text to grade-8 reading level.

Rewrite the text below in plain English at a grade-{{grade_level}} reading level.

Rules:
- Keep every fact
- Sentences ≤ 20 words
- No jargon unless you define it inline
- Keep th…
writingGPT-4oGeminiClaude{{2}}#editing#plain-language
11 uses · Aug 26, 2026Open →

Extracts typed fields from messy text into strict JSON.

Extract the following fields from the text below and return ONLY valid JSON (no markdown fence):

{
  "person": string | null,
  "company": string | null,
  "date": "YYYY-MM-DD" | …
researchGPT-4oClaude{{1}}#extraction#json
15 uses · Aug 26, 2026Open →

Senior-engineer code review: correctness, security, style — ranked by severity.

You are a staff engineer reviewing a pull request in {{language}}.

Review the code below for:
1. Correctness bugs (rank P0/P1/P2)
2. Security issues
3. Readability and naming
4. M…
codingClaudeGPT-4.1{{2}}#code-review#engineering
21 uses · Aug 26, 2026Open →

Condenses any long text for a specific audience at a target length.

Summarize the following {{topic}} for a {{audience}} in {{length}} words.

Focus on what matters to that audience. Use plain language and short paragraphs.

TEXT:
"""
{{input}}
"""
writingGPT-4oClaude{{4}}#general#summarization
12 uses · Aug 26, 2026Open →

Asks the model to parse logs into structured rows.

Parse the {{log_type}} log lines below into a markdown table with columns: {{timestamp}} | level | component | message | trace_id (if present).

Drop health-check noise lines (cont…
analysisGPT-4o{{3}}#logs#parsing
0 uses · Aug 26, 2026Open →

Translates text while respecting a term glossary.

Translate the text below from {{source_language}} to {{target_language}}.

Glossary (never translate these terms):
{{glossary}}

Tone: {{tone}}. Preserve formatting and line breaks…
writingGPT-4oGemini{{4}}#localization#translation
2 uses · Aug 26, 2026Open →

Explains a regex in English, then simplifies it.

Explain what this regular expression matches, piece by piece:

/{{regex}}/

Then: (a) write a plain-English one-liner, (b) if it can be simplified without changing behavior, show t…
codingClaudeGPT-4o{{1}}#explainer#regex
5 uses · Aug 26, 2026Open →

Five hero headline+subhead pairs ranked by clarity.

Write 5 hero-section pairs (headline ≤ 10 words, subhead ≤ 25 words) for {{product}}.

Who it's for: {{audience}}
Primary pain: {{pain}}
Differentiator: {{wedge}}

Rank them by cla…
marketingGPT-4oGemini{{4}}#copywriting#landing-page
4 uses · Aug 26, 2026Open →

Blameless postmortem structure filled from raw timeline.

Draft a blameless postmortem for incident "{{incident}}".

Raw timeline:
{{timeline}}

Sections: Summary · Impact (users/duration) · Timeline · Root cause · What went well · What w…
analysisGPT-4o{{2}}#incident#sre
1 use · Aug 26, 2026Open →

Locks a persona's voice for long-form fiction.

Maintain this character sheet while writing {{scene}}.

CHARACTER:
{{character}}

Rules:
- Never break the voice defined above
- Keep responses ≤ {{length}} words
- End every scene…
writingClaudeGPT-4o{{3}}#fiction#persona
2 uses · Aug 26, 2026Open →

Turns a git log into user-facing release notes.

Turn the commit log below into release notes for {{product}} v{{version}}.

Audience: {{audience}}

Group into: ✨ New · 🐛 Fixed · ⚡ Improved. Skip internal refactors. One line per…
writingGPT-4oClaude{{3}}#changelog#developer-relations
6 uses · Aug 26, 2026Open →

Builds a prioritized learning path with time estimates.

Build a {{duration}} learning path for {{topic}}.

My background: {{background}}
Goal: {{goal}}

Format: ordered list. Each item: resource type, why this order, and estimated hours…
researchGeminiGPT-4o{{4}}#curation#learning
3 uses · Aug 26, 2026Open →

Critiques a REST API contract for consistency and DX.

Review this REST API contract for developer experience:

{{contract}}

Check: naming consistency, status-code discipline, pagination, error envelope, versioning, and idempotency of…
codingClaude{{1}}#api#review
1 use · Aug 26, 2026Open →

3-sentence cold outreach with a specific trigger.

Write a cold email to a {{role}} at a {{company_type}} company.

Offer: {{offer}}
Trigger observation: {{trigger}}

Rules: exactly 3 sentences, no "I hope this finds you well", one…
marketingGPT-4o{{4}}#email#sales
2 uses · Aug 26, 2026Open →

Converts raw transcript into decisions + owned action items.

Convert the meeting transcript below into:

1. DECISIONS (bullet list, one line each)
2. ACTION ITEMS (table: task | owner | due)
3. OPEN QUESTIONS (unresolved threads)

If an owne…
analysisClaudeGPT-4o{{1}}#meetings#productivity
8 uses · Aug 26, 2026Open →

Translates a natural-language question into a safe SELECT.

Given this schema:

{{schema}}

Write a single PostgreSQL query answering: "{{question}}"

Rules: SELECT statements only, use CTEs for readability, comment each CTE in one line.
codingGPT-4.1{{2}}#data#sql
5 uses · Aug 26, 2026Open →

Structured competitor analysis from their marketing site.

Analyze the competitor "{{competitor}}" in the {{industry}} market.

Based on the material below, produce:
1. Positioning statement (their one-liner)
2. Pricing strategy inference
…
analysisClaudeGPT-4o{{3}}#research#strategy
6 uses · Aug 26, 2026Open →

Structured image-gen prompt with camera and lighting language.

A professional product photograph of {{product}}, on a {{surface}} surface with a {{background}} background, {{lighting}} lighting, shot at {{angle}}, 85mm lens, f/2.8, high detail…
imageGeminiOther{{5}}#ecommerce#image
2 uses · Aug 26, 2026Open →

Generates table-driven unit tests for a function.

Write {{framework}} unit tests for the function below.

Requirements:
- Table-driven style (cases array)
- Cover: happy path, boundary values, empty input, unicode, and one adversa…
codingGPT-4.1Claude{{2}}#code-generation#testing
9 uses · Aug 26, 2026Open →

Generates ad variants across hooks (pain, aspiration, proof, FOMO).

Write {{count}} ad copy variants for {{product}}.

Target: {{audience}}
Platform: {{platform}}
Brand voice: {{tone}}
Constraint: max {{char_limit}} characters per variant.

Use fou…
marketingGPT-4oGemini{{6}}#ads#copywriting
3 uses · Aug 26, 2026Open →

Base system prompt for a no-fluff terminal assistant.

You are a Linux systems expert. Answer with the exact command or config first, then at most 2 lines of explanation. Never use GUI instructions. Assume Debian-based distros unless t…
systemClaudeLlamaMistral#persona#system-prompt
4 uses · Aug 26, 2026Open →

SEO-aware outline with H2/H3 structure and hook angles.

Create a blog post outline for the topic "{{topic}}".

Audience: {{audience}}
Tone: {{tone}}
Target length: {{length}} words

Return:
- 5 candidate titles (one curiosity-gap, one h…
marketingGPT-4oGemini{{4}}#blog#content#seo
7 uses · Aug 26, 2026Open →