Java AI Prompt Testing Framework

A PromptOps workspace for testing, versioning, comparing, and governing LLM prompts.

Product direction

The project is evolving from a developer utility into an enterprise-ready prompt experimentation platform:

Prompt portfolio dashboard with readiness and challenger coverage metrics.
Version-controlled prompt library with comparison workflows.
Quick playground for local evaluation and demo-safe mock mode.
Governance queue for prompts missing owner context, challenger versions, or recent review.
Release governance gate with publishable counts, blockers, required checks, and rollout risk disclosure.
JSON/API-friendly frontend data model designed to connect to the Spring backend.

The Spring backend now exposes the PromptOps workspace contract used by the frontend:

GET /api/workspace/summary returns portfolio readiness, challenger coverage, release governance, and governance queue rows.
Prompt, prompt-version, quick-test, and test-run APIs back the core experimentation flow.
The workspace summary mirrors the frontend mock analytics rules so demo and live modes use the same product semantics.
Release governance uses PromptOps.ReleaseGovernance.v1 and blocks rollout when prompts lack versions, owner context, challenger versions, or recent review.

Planned backend expansion remains focused on CI-style gates for prompt quality, latency, cost, and regression thresholds.

Prerequisites:

cd frontend
npm install
npm run dev

Mock mode is enabled by default for local demos. Set VITE_USE_MOCK=false and VITE_API_BASE_URL to connect to a backend.

cd frontend
npm run test:all
npm run build