# Prompt‑Vault™

## Overview
Prompt‑Vault™ is an all‑in‑one workspace for AI prompt engineers, marketers, and developers. It centralises:

* **Searchable Prompt Library** – tag‑rich, categorised prompts ready for reuse.  
* **Immutable Version‑Control Log** – every edit is recorded with a cryptographic hash to guarantee provenance.  
* **Integrated Test‑Run Tracker** – run prompts against models, capture outputs, metrics, and notes.  
* **Asset Store** – store prompt‑related files (datasets, images, scripts) with versioned links.  
* **Deployment Checklist** – a repeatable, gated process to move prompts from dev to production.

The workflow is linear but flexible:

1. **Create / import** a prompt in the Prompt Library.  
2. **Commit** the initial version – an entry appears in the Version Control Log.  
3. **Run tests** in the Test Run Tracker; iterate and commit new versions as needed.  
4. **Attach assets** (e.g., fine‑tuning data) in the Asset Store and link them.  
5. **Complete the Deployment Checklist** to promote the prompt to production.

> 💡 Tip: Use the global search (⌘ K) to find prompts by tag, model, or author instantly.

---

## Prompt Library
| Prompt ID | Title | Category | Model | Tags | Owner | Last Updated |
|-----------|-------|----------|-------|------|-------|--------------|
| PRM‑001 | SEO Meta Description Generator | Marketing | gpt‑4o | seo, meta, copywriting | Alice | 2024‑11‑02 |
| PRM‑012 | Code Review Summarizer | Development | Claude‑3.5 | code, review, summarise | Bob | 2024‑10‑28 |
| PRM‑023 | Customer Support Tone Adjuster | Support | gpt‑4o-mini | support, tone, empathy | Carla | 2024‑11‑05 |
| PRM‑037 | Data‑Extraction Prompt (Invoices) | Data Ops | Gemini‑1.5 | extraction, finance, regex | Dan | 2024‑09‑30 |
| PRM‑045 | Creative Story Starter | Creative | gpt‑4o | story, creative, writing | Eve | 2024‑11‑01 |

> 💡 Tip: Click a Prompt ID to open its **Prompt Detail** page, where you can view the full prompt text, version history, and linked assets.

---

## Version Control Log (Immutable)
| Commit Hash | Prompt ID | Version | Author | Timestamp (UTC) | Change Summary |
|-------------|-----------|---------|--------|-----------------|----------------|
| `a3f9c2e7` | PRM‑001 | v1.0 | Alice | 2024‑11‑02 08:12 | Initial creation |
| `d5b1f4a9` | PRM‑001 | v1.1 | Alice | 2024‑11‑03 14:45 | Added temperature=0.7, refined tone |
| `9e2c7b3d` | PRM‑012 | v1.0 | Bob | 2024‑10‑28 09:30 | First draft for code review |
| `4f8a6d1c` | PRM‑023 | v1.0 | Carla | 2024‑11‑05 11:02 | Baseline prompt for support |
| `c0e7b9f2` | PRM‑037 | v2.0 | Dan | 2024‑10‑15 16:20 | Updated regex for new invoice format |

> 💡 Tip: The hash is generated from the prompt content + metadata; any alteration creates a new row, preserving the original record forever.

---

## Test Run Tracker
| Test ID | Prompt ID | Model | Input Sample | Output Snapshot | Success Metric | Tester | Date |
|---------|-----------|-------|--------------|----------------|----------------|--------|------|
| TST‑101 | PRM‑001 | gpt‑4o | “Eco‑friendly travel blog” | “Explore sustainable destinations…” | BLEU = 0.78 | Alice | 2024‑11‑04 |
| TST‑112 | PRM‑012 | Claude‑3.5 | Pull request #42 diff | “Key changes: …” | Accuracy = 92% | Bob | 2024‑10‑29 |
| TST‑127 | PRM‑023 | gpt‑4o-mini | Customer complaint email | “We’re sorry …” | Sentiment = Positive | Carla | 2024‑11‑06 |
| TST‑138 | PRM‑037 | Gemini‑1.5 | Invoice PDF (sample) | “Total: $1,245.00” | Extraction = 98% | Dan | 2024‑10‑20 |

> 💡 Tip: Attach the full raw output as a file in the **Asset Store** and link it back to the test row for auditability.

---

## Asset Store
| Asset ID | Name | Type | Linked Prompt(s) | Version | Uploaded By | Date |
|----------|------|------|------------------|---------|-------------|------|
| AST‑001 | SEO Prompt Template.docx | Document | PRM‑001 | v1 | Alice | 2024‑11‑02 |
| AST‑014 | Code Review Sample.diff | Text | PRM‑012 | v1 | Bob | 2024‑10‑28 |
| AST‑022 | Support Tone Guidelines.pdf | PDF | PRM‑023 | v2 | Carla | 2024‑11‑05 |
| AST‑035 | Invoice Sample Set.zip | Archive | PRM‑037 | v3 | Dan | 2024‑09‑30 |
| AST‑047 | Story Starter Prompts.csv | CSV | PRM‑045 | v1 | Eve | 2024‑11‑01 |

> 💡 Tip: Use the **Relation** field “Linked Prompt(s)” to automatically surface assets on a prompt’s detail page.

---

## Deployment Checklist
| Step | Description | Owner | Status | Completed On |
|------|-------------|-------|--------|--------------|
| 1️⃣ Review Prompt Version | Verify latest version is approved by peer review. | Prompt Owner | ✅ | 2024‑11‑06 |
| 2️⃣ Run Production Test | Execute test with live model endpoint; capture metrics. | QA Engineer | ⬜ | — |
| 3️⃣ Attach Final Assets | Link final asset package (e.g., fine‑tuned model, data). | Asset Manager | ⬜ | — |
| 4️⃣ Update Documentation | Ensure README and usage docs reflect new prompt. | Technical Writer | ⬜ | — |
| 5️⃣ Publish to Production DB | Move prompt entry to “Production” view; lock editing. | Ops Lead | ⬜ | — |
| 6️⃣ Notify Stakeholders | Send release note via Slack/email. | Release Manager | ⬜ | — |

> 💡 Tip: The checklist is a **Kanban board** view in Notion; drag cards to “Done” to automatically set the ✅ status.

---

## How to import into Notion
1. **Create a new page** in your Notion workspace where you want Prompt‑Vault™.  
2. Click **Import** → **Markdown & CSV**, then **upload this .md file**. Notion will generate a page with all headings and tables.  
3. For each table (Prompt Library, Version Control Log, etc.):  
   a. Click the three‑dot menu on the table → **Turn into → Database – Full page** (or **Inline** if you prefer).  
   b. Verify column types (e.g., set “Timestamp” to **Date**, “Success Metric” to **Number**, “Tags” to **Multi‑select**, “Linked Prompt(s)” to **Relation**).  
4. Re‑create the **Relation** fields:  
   * In the Prompt Library, add a Relation to **Version Control Log** (by Prompt ID).  
   * In the Asset Store, add a Relation to **Prompt Library**.  
   * In the Test Run Tracker, add Relations to **Prompt Library** and **Asset Store** as needed.  
5. Add the **Kanban view** for the Deployment Checklist: click **Add a view**, select **Board**, group by **Status**.  
6. (Optional) Set up **Filters** on the Prompt Library view to show only “Production” prompts, or on the Test Run Tracker to display the last 7 days.  
7. Share the page with your team, assign appropriate permissions, and start populating!