Skip to main content
A project is the boundary for a single AI application in HoneyHive. All traces, datasets, evaluations, prompts, and metrics live within a project. Each project has its own API key for SDK authentication. Projects live inside workspaces. New organizations include a starter workspace and project so you can create an API key and begin tracing immediately. To create another project, navigate to your workspace and click New Project. Project names can only contain letters, numbers, and underscores, and must be unique within a workspace. To issue credentials from a provisioning service after creating a project, see How to create API keys programmatically.

​
Organizing projects

For complex AI pipelines, we recommend creating multiple projects within a workspace:
  1. A production project (e.g., “Chatbot Production”)
    • Source of truth for production traces and CI tests
    • Keep schemas stable and aligned with what runs in production
  2. One project per testable component (e.g., “Chatbot Retriever”)
    • For offline evaluation, prompt management, and experiments
    • Track what worked for a specific piece of the pipeline
  3. A catch-all project (e.g., “Chatbot Random”)
    • For ad-hoc experiments and ideas worth keeping but not actively referenced

​
Archiving projects

Archive a project when it is no longer actively used. Archiving hides the project and its data without permanently deleting them.

​
Access control

Project access is managed independently from workspace access. A workspace membership alone does not grant access to project data. Add users to the project with a data-granting role so they can see its data. Parent-scope admin roles inherit scope and membership management on descendant projects. See Permission inheritance.
  • Project Admins have full control over the project and its members.
  • Project Members can view and create most resources but cannot delete them or manage membership roles.
See Roles for the full permission breakdown.