Sprint 2 — Goal
Make AgentOps Toolkit easier to adopt directly from developer workflows through GitHub Copilot support.
Sprint 2 — User Experience Summary
How a developer uses AgentOps Toolkit
A developer installs AgentOps Toolkit from PyPI and runs a simple initialization command inside their own agent project. This adds the basic AgentOps files needed to start using the toolkit from within their repository without changing the main structure of the project. From there, the developer can use the AgentOps CLI as part of their normal development workflow.
How GitHub Copilot uses AgentOps Toolkit
When AgentOps Toolkit is initialized in a project, it can also add lightweight Copilot guidance files to that same repository. This allows GitHub Copilot, whether in VS Code or Copilot CLI, to understand that the project uses AgentOps Toolkit and to suggest or run the correct commands accordingly. Instead of relying on ad hoc scripts, Copilot can guide developers to use the AgentOps CLI consistently for setup and evaluation workflows.
Scope
- Enable AgentOps Toolkit to be installed as a Python CLI and initialized inside any agent development repository
- Add an
agentops init experience that scaffolds the minimum files needed to use the toolkit in a target project
- Add lightweight GitHub Copilot guidance files to the target repository without overwriting existing user instructions
- Support GitHub Copilot usage in both VS Code and Copilot CLI
- Focus the Copilot experience on Foundry agent evaluations
- Validate Sprint 1 foundations through a focused bug bash and capture adoption issues
Deliverables
-
agentops init command to bootstrap AgentOps Toolkit into a target agent project
-
Initial project scaffolding for AgentOps configuration and evaluation usage
-
Repo-local Copilot guidance files added safely as part of initialization
-
Documentation explaining:
- how developers install and initialize AgentOps Toolkit
- how GitHub Copilot uses the toolkit in VS Code and Copilot CLI
-
Sprint 1 bug bash findings captured as GitHub issues
Definition of Done
- Developers can install AgentOps Toolkit with minimal setup
- Developers can run
agentops init inside their own repository and get the required starter files
- GitHub Copilot can use AgentOps Toolkit guidance from the target repository in both VS Code and Copilot CLI
- Existing user Copilot instructions are not overwritten by default
- Documentation explains setup and Copilot usage clearly
- Sprint 1 bug bash is completed and findings are logged for follow-up
Sprint 2 — Goal
Make AgentOps Toolkit easier to adopt directly from developer workflows through GitHub Copilot support.
Sprint 2 — User Experience Summary
How a developer uses AgentOps Toolkit
A developer installs AgentOps Toolkit from PyPI and runs a simple initialization command inside their own agent project. This adds the basic AgentOps files needed to start using the toolkit from within their repository without changing the main structure of the project. From there, the developer can use the AgentOps CLI as part of their normal development workflow.
How GitHub Copilot uses AgentOps Toolkit
When AgentOps Toolkit is initialized in a project, it can also add lightweight Copilot guidance files to that same repository. This allows GitHub Copilot, whether in VS Code or Copilot CLI, to understand that the project uses AgentOps Toolkit and to suggest or run the correct commands accordingly. Instead of relying on ad hoc scripts, Copilot can guide developers to use the AgentOps CLI consistently for setup and evaluation workflows.
Scope
agentops initexperience that scaffolds the minimum files needed to use the toolkit in a target projectDeliverables
agentops initcommand to bootstrap AgentOps Toolkit into a target agent projectInitial project scaffolding for AgentOps configuration and evaluation usage
Repo-local Copilot guidance files added safely as part of initialization
Documentation explaining:
Sprint 1 bug bash findings captured as GitHub issues
Definition of Done
agentops initinside their own repository and get the required starter files