Constitution vs AGENTS.md vs SKILLS #2268
Replies: 3 comments
-
|
Any and all documents have their place. As the LLM can inspect any of them at their own discretion they can all influence the process from that perspective. However from the perspective of each of the Spec KIt commands the constitution is part of that and to a lesser degree AGENTS.md (dependent on the coding agent you enable) and skills are outside of the purview of the Spec Kit commands with the exception of the implement command. When I say purview it means there is no direct mention |
Beta Was this translation helpful? Give feedback.
-
|
I'm having a hard time understanding why
If the harness is in control of most things, are you highjacking that? Or should it listen to both? For instance, should we not use |
Beta Was this translation helpful? Give feedback.
-
|
We have made some changes recently to address this. It now uses a guarded block to keep the changes limited to just one cycle of the Spec Kit SDD process |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
We are new to spec-kit and looking for clarification on when to use Constitution, AGENTS.md, and SKILLS in spec‑kit, and how to decide what type of content belongs in each:
What is the intended responsibility and scope of each artifact (Constitution, AGENTS.md, and SKILLS)?
How should we decide what content goes where to avoid overlap or duplication?
Are there clear guidelines or examples that differentiate architectural/design rules from agent behavior and skills?
For brownfield projects, is defining SKILLS actually required, or is it optional when existing systems and patterns already exist?
If Constitution is created, do we will need to add the Architecture document for spec-kit to work correctly?
Any best practices or real‑world examples would be greatly appreciated.
Thanks.
Beta Was this translation helpful? Give feedback.
All reactions