AgentUX
Portable Multi-Agent UX Stress-Tester for local web apps.
Goal
Install once, then run AgentUX from any project root to audit that project's running app.
Install
pip install -e .
playwright install chromium
Install Anywhere
You have three practical options:
- Direct from local path (dev mode)
pip install -e /absolute/path/to/ux-tester
- Isolated global CLI via
pipx(recommended for tool-style usage)
pipx install /absolute/path/to/ux-tester
- Build wheel and install on another machine
python -m pip install build
python -m build
pip install dist/agentux-0.1.0-py3-none-any.whl
Use In Any Project
From the target project's root directory:
- Install AgentUX
pip install agentux
- Initialize AgentUX (sets key + infers project context/scenario)
agentux init
- Start your web app (framework-specific)
# examples:
npm run dev
# or
pnpm dev
# or
python manage.py runserver
- Run AgentUX
agentux run
agentux run:
- Requires
.agentux/to already exist (created byagentux init) - Ensures Playwright Chromium is installed (unless
--skip-browser-install) - Starts dashboard and auto-starts audit
Optional:
agentux run --no-auto
Run:
agentux run
Open http://localhost:8501.
Features
- Discovery + dynamic persona synthesis
- Parallel persona audit loops with Playwright
- Structured observability artifacts (screenshots, traces, JSONL logs)
- Streamlit mission-control dashboard
- Final UX friction report (
report.md+report.json)
Artifact Output
Each run writes under the current project's .agentux:
.agentux/artifacts/<run_id>/
config_snapshot.json
discovery/
agents/<agent_id>/
aggregate/
Scenario config is read from .agentux/scenarios/default.yaml.
Login credentials are read on-demand from .agentux/login.yaml when agents need to fill auth forms.
Use this format:
accounts:
- username: student@example.com
password: your-password
- username: teacher@example.com
password: your-password
Inference details are saved to .agentux/project_profile.json.
Example:
default_scenario:
id: "default"
name: "Checkout Flow"
goal: "Complete a purchase"
success_assertions:
- type: "url_contains"
value: "/checkout/complete"
persona_overrides: {}
Optional Docker Dev Mode
docker-compose up --build
Notes
agentux initupdates.gitignorewith.agentux/.env,.agentux/login.yaml, and.agentux/artifacts/.- The engine fails fast on missing API key or unreachable target URL.
- Your target app must already be running before
agentux run.
Metadata
Release files for agentux 0.1.4
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentux-0.1.4.tar.gz | 30.6 kB | Without provenance | Details for agentux-0.1.4.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentux-0.1.4-py3-none-any.whl | Python 3 | none | any | Details for agentux-0.1.4-py3-none-any.whl |
Total release size: 62.8 kB
Release files / agentux-0.1.4.tar.gz
| Download URL | agentux-0.1.4.tar.gz |
|---|---|
| Size | 30.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8993b74bf783c6d3c6dc43d13455821181853ea4d7d7a21fb1a776cd2e73baf6
|
|
BLAKE2b-256 checksum How to use checksums |
3252f4aaf37ba2dc40e836dfe43d82bc0f44e5fb62c9e62a17bb6174deeea7a5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.7
|
Release files / agentux-0.1.4-py3-none-any.whl
| Download URL | agentux-0.1.4-py3-none-any.whl |
|---|---|
| Size | 32.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
597562bfc29d741d57af9bd8e3a6906cbd26f4f9f9b9f42dcefd8bcc5b3a21f5
|
|
BLAKE2b-256 checksum How to use checksums |
66dc76005262f7efe0476880026340bcc7cca4a823327095c46477657b5ee67b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.7
|
Security for agentux 0.1.4
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.