AppVA¶

AppVA is an automated tool including two parts:

1. A pipeline that converts existing Espresso UI tests (Java/Kotlin) into VA methods and using LLM calls (GPT APIs) to generate other code description artifacts.
2. A runtime server and client architecture that support converting user request to the VA tasks.

Demo video¶

You can watch a short demo of AppVA in action here:

https://youtu.be/z4p19QL6ejw

Quick start¶

1. Install the package from the project root:

pip install .

This installs the appva CLI and the runtime components.

1. Configure environment and OpenAI:
2. Copy .env.example to .env, and edit it

cp .env.example .env

• Set OPENAI_API_KEY and (optionally) variables such as AVA_GEN_OPENAI_MODEL

• Run your first end-to-end example using the real workspace/ folder and the bundled app hu.vmiklos.plees_tracker by following:

• docs/running_example.md

This walks you through preparing inputs under workspace/, running appva pipeline hu.vmiklos.plees_tracker, and inspecting the generated artifacts (extracted tests, VA methods, skills, intents, action plans).

1. Explore the CLI and pipelines for your own app (see docs/cli.md and docs/getting-started.md for details):

appva --help

1. Start the runtime server (once workspace artifacts and action plans exist):

uvicorn runtime.api.server:app --reload

See the docs/ directory (used by MkDocs) for:

• docs/running_example.md – concrete running example on the real workspace.
• docs/cli.md – CLI usage and pipeline overview.
• docs/runtime_server_client.md – runtime usage.