Skip to content

Latest commit

 

History

History
225 lines (156 loc) · 8.79 KB

feature-request.md

File metadata and controls

225 lines (156 loc) · 8.79 KB

🖥️ AI Driven Dev - "Feature request" Prompts (:feat)

This document provides you with a list of prompts that you can use to request the AI to generate new features or to help you implement existing ones.

🆕 Create new features

Create a short feature in the code :featCodeSpec

Description:

Create a short feature in the code, from the IDE, interacting directly in the codebase.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool GitHub Copilot
🔍 Context Integrate a small feature rapidly in the codebase.

Prompt:

@workspace from #codebase I need to implement this new feature:

---
[[Feature to build, be as detailed as possible]]
---

List all the required steps to code it.

Generate feature as User-Stories :featGenerateUS

Description:

From a feature, generate a full user-stories with acceptance criteria and sub-tasks.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool ChatGPT, GPT Agile Coach
🔍 Context If you feature is clear enough to generate a US.

Prompt:

Goal: Generate a concise but full breakdown of this feature for our project.

Rules:
- Look for that information in your knowledge base to provide the best answer.
- If user-story is too big, group features the more you can.
- You MUST provide every required steps, do not forget anything.
- Do not generate code or command line instructions, focus on precise steps to achieve the outcome.
- It should include epics, user stories with acceptance criteria, sub-tasks.

Requested Feature (delimited by "---"):
---
[[Feature to build, be as detailed as possible]]
---

[[Project and lib setup have already been done | Project and lib setup might not have been done]].

Here is a styled markdown template example you can inspire yourself with (surrounded by "---" delimiter):
---
# Feature's name with Epic

## "User Story 1"

* Acceptance Criteria:
  * Given: ...
  * When: ...
  * Then: ...
  * And: ...

### Sub-tasks

* "Sub-task title 1.1"
  * Full step to achieve the outcome
  * ...
---

Extract development steps :featExtractDevSteps

Description:

From a feature, extract the development steps for each sub-task.

Ask the LLM to generate fully detailed ticket based on the user stories for an individual sub-task.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool ChatGPT, GPT RAG Dev
🔍 Context If the user-story is clear.
✨ Required before Sub-task of a User-Story required.
✅ External files Re-upload project-documentation.txt first.

Prompt:

Given the task detailed below (surrounded by "---" delimiters), can you provide me with the development steps for each sub-task?

---
[[sub task full description]]
---

1. For each question, search into knowledge base.
2. Recall name, tech stack and libraries from [[package.json | composer.json | requirements.txt]] in the knowledge base.
3. Concentrate solely on the task at hand.
4. Identify any potential missing sub-tasks.
5. Reformulate tasks in one line.
6. Look for that information in your `project-documentation.txt` file in your knowledge base to provide the best answer.
7. Provide one-line action steps, specifying new or existing file paths without generating code.

Extract coding steps :featExtractCodingSteps

Description:

Base on the detailed development steps, extract the coding steps for each sub-task.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool ChatGPT, GPT RAG Dev
🔍 Context Go to development steps.
✨ Required before Development steps extracted before.
✅ External files Uploaded project-documentation.txt.

Prompt:

Recalling our earlier discussion about the project tasks:

[[List of development steps if not specified above]]

1. For required libraries, list their versions for code generation.
2. Examine each sub-task for prerequisites or missing configurations, providing details as needed.
3. Consult the project structure information and advise on whether to create new files or modify existing ones.
4. Outline detailed coding steps.
5. Produce code using the most recent tech versions from our discussions, adhering to best practices.

👌 Existing features

How to implement :featExistingImplement

Description:

For an existing feature, extract the development steps for each sub-task.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool ChatGPT, GPT RAG Dev
🔍 Context If you need to implement a new feature.

Prompt:

Here are the specifications of the feature I need to code (surrounded by "---" delimiters).
---
[[Copy and paste the specifications here]]
---

Now, can you:
- Acknowledge it.
- Reformulate in bullet point grouped by section to show me that you understood what to do.
- Generate development steps (based on an existing projet you do not know nothing about).

Finally, ask me the relevant questions about implementing the feature in my project regarding my project's tech stack and libraries.

Ask anything you need to know, like existing code, libraries, mockups available, logics, etc.

How to implement (answer) :featExistingImplementAnswer

Description:

Answer the Chat about how to implement the existing feature.

How to use the prompt:

Usage Context
🚀 Author @alexsoyes
🤖 Tool ChatGPT, GPT RAG Dev
🔍 Context After you started to talk to implement the existing feature.
✨ Required before How to implement prompt

Prompt:

Here are some answers about your previous questions.

Note that I cannot provide you with the full code, nor answer every details of the question.

Can you merge this new information together with the previous ones, and then:

- If there is not absolute need for more details, only give me the required development steps involved in coding the functionality?
- If you definitely need more details, please only ask me the crippling questions.

Here are my answers:

[[Your answers to the previous questions]]