id: edge-ai-tasks
name: Tasks by microsoft/edge-ai
description: Task Researcher and Task Planner for intermediate to expert users and large codebases - Brought to you by microsoft/edge-ai
tags: [architecture, planning, research, tasks, implementation]
items:
Planning Chat Modes
-
path: agents/task-researcher.agent.md
kind: agent
usage: |
Now you can iterate on research for your tasks!
---
mode: task-researcher
title: Research microsoft fabric realtime intelligence terraform support
---
Review the microsoft documentation for fabric realtime intelligence
and come up with ideas on how to implement this support into our terraform components.
Research is dumped out into a .copilot-tracking/research/*-research.md file and will include discoveries for GHCP along with examples and schema that will be useful during implementation.
Also, task-researcher will provide additional ideas for implementation which you can work with GitHub Copilot on selecting the right one to focus on.
-
path: agents/task-planner.agent.md
kind: agent
usage: |
Also, task-researcher will provide additional ideas for implementation which you can work with GitHub Copilot on selecting the right one to focus on.
---
mode: task-planner
title: Plan microsoft fabric realtime intelligence terraform support
---
#file: .copilot-tracking/research/*-fabric-rti-blueprint-modification-research.md
Build a plan to support adding fabric rti to this project
task-planner will help you create a plan for implementing your task(s). It will use your fully researched ideas or build new research if not already provided.
task-planner will produce three (3) files that will be used by task-implementation.instructions.md.
-
.copilot-tracking/plan/*-plan.instructions.md
- A newly generated instructions file that has the plan as a checklist of Phases and Tasks.
-
.copilot-tracking/details/*-details.md
- The details for the implementation, the plan file refers to this file for specific details (important if you have a big plan).
-
.copilot-tracking/prompts/implement-*.prompt.md
- A newly generated prompt file that will create a
.copilot-tracking/changes/*-changes.md file and proceed to implement the changes.
Continue to use task-planner to iterate on the plan until you have exactly what you want done to your codebase.
Planning Instructions
-
path: instructions/task-implementation.instructions.md
kind: instruction
usage: |
Continue to use task-planner to iterate on the plan until you have exactly what you want done to your codebase.
When you are ready to implement the plan, create a new chat and switch to Agent mode then fire off the newly generated prompt.
---
mode: agent
title: Implement microsoft fabric realtime intelligence terraform support
---
/implement-fabric-rti-blueprint-modification phaseStop=true
This prompt has the added benefit of attaching the plan as instructions, which helps with keeping the plan in context throughout the whole conversation.
Expert Warning ->>Use phaseStop=false to have Copilot implement the whole plan without stopping. Additionally, you can use taskStop=true to have Copilot stop after every Task implementation for finer detail control.
To use these generated instructions and prompts, you'll need to update your settings.json accordingly:
"chat.instructionsFilesLocations": {
// Existing instructions folders...
".copilot-tracking/plans": true
},
"chat.promptFilesLocations": {
// Existing prompts folders...
".copilot-tracking/prompts": true
},
display:
id: edge-ai-tasks
name: Tasks by microsoft/edge-ai
description: Task Researcher and Task Planner for intermediate to expert users and large codebases - Brought to you by microsoft/edge-ai
tags: [architecture, planning, research, tasks, implementation]
items:
Planning Chat Modes
path: agents/task-researcher.agent.md
kind: agent
usage: |
Now you can iterate on research for your tasks!
Research is dumped out into a .copilot-tracking/research/*-research.md file and will include discoveries for GHCP along with examples and schema that will be useful during implementation.
Also, task-researcher will provide additional ideas for implementation which you can work with GitHub Copilot on selecting the right one to focus on.
path: agents/task-planner.agent.md
kind: agent
usage: |
Also, task-researcher will provide additional ideas for implementation which you can work with GitHub Copilot on selecting the right one to focus on.
task-plannerwill help you create a plan for implementing your task(s). It will use your fully researched ideas or build new research if not already provided.task-plannerwill produce three (3) files that will be used bytask-implementation.instructions.md..copilot-tracking/plan/*-plan.instructions.md.copilot-tracking/details/*-details.md.copilot-tracking/prompts/implement-*.prompt.md.copilot-tracking/changes/*-changes.mdfile and proceed to implement the changes.Continue to use
task-plannerto iterate on the plan until you have exactly what you want done to your codebase.Planning Instructions
path: instructions/task-implementation.instructions.md
kind: instruction
usage: |
Continue to use
task-plannerto iterate on the plan until you have exactly what you want done to your codebase.When you are ready to implement the plan, create a new chat and switch to
Agentmode then fire off the newly generated prompt.This prompt has the added benefit of attaching the plan as instructions, which helps with keeping the plan in context throughout the whole conversation.
Expert Warning ->>Use
phaseStop=falseto have Copilot implement the whole plan without stopping. Additionally, you can usetaskStop=trueto have Copilot stop after every Task implementation for finer detail control.To use these generated instructions and prompts, you'll need to update your
settings.jsonaccordingly:display: